Multimodal large language models (MLLMs) have achieved remarkable progress in visual understanding tasks. However, most existing MLLMs rely on autoregressive generation, which limits their efficiency for perception tasks that require captioning multiple regions. In this work, we propose PerceptionDLM, a multimodal diffusion language model optimized for efficient parallel region perception. Built upon PerceptionDLM-Base, a strong foundational baseline that achieves state-of-the-art performance among open-source diffusion MLLMs, our architecture fully leverages the parallel decoding nature of DLMs. Specifically, we introduce efficient prompting and structured attention masking to enable simultaneous perception of multiple masked regions, allowing the model to generate region descriptions in parallel at both the sequence and token levels. This design significantly improves inference efficiency compared with existing approaches that process regions sequentially. To systematically evaluate the parallelism property of visual perception capability for DLMs, we construct a new Parallel Detailed Localized Captioning Benchmark (ParaDLC-Bench) by scaling the DLC-Bench to include multiple region masks per image, enabling joint evaluation of both caption quality and inference efficiency. Experiments demonstrate that PerceptionDLM maintains competitive performance in region captioning while achieving substantial speed improvements for multi-region perception tasks. Our results highlight the potential of multimodal diffusion language models for efficient, parallel visual perception. To the best of our knowledge, we are the first to achieve parallel region caption and perception by leveraging the advantages of diffusion language models. Code, models, and datasets are released.

Vision foundation models are typically trained as static feature extractors, placing the burden of task adaptation onto large downstream models. We propose an alternative paradigm: instead of solely feeding visual features into language models, we use language itself to dynamically guide the vision encoder. Our method, Language-Instructed Vision Embeddings (LIVE), leverages language as high-level guidance to produce task-centric embeddings at inference time, removing the need for task-specific retraining. This enables the encoder to focus on contextually relevant aspects of the input, yielding more controllable and generalizable representations. Empirically, LIVE reduces visual hallucinations (+34 points on MMVP), surpasses vision-language models with orders of magnitude more parameters on visual question answering, and generalizes to unseen instructions and tasks -- offering a direct path toward adaptive, instruction-driven visual intelligence.

3D multimodal large language models (3D MLLMs) describe a 3D object as a whole but cannot address, name, or reason about its parts. Prior part-aware attempts add segmentation decoders, heavier 3D encoders, or bounding-box grammars at substantial parameter cost. We take a fundamentally different path: we reorganize the input token stream so that parts become directly addressable through the LLM's own vocabulary. Our model, 3D-PLOT-LLM, partitions the frozen point encoder's patches into K locally coherent regions and inserts, before each region's patch tokens, a learnable per-region marker and a reserved vocabulary token <part_k>; a Marker-Space Refinement (MSR) module then conditions each marker on its region's spatial statistics and adjacency neighbors. The model thus cites parts in its output and follows prompts that refer to parts by token, a capability absent from prior object-level 3D MLLMs. To probe this interface, we construct PartVerse-QA, a vocabulary-level part-QA benchmark adapted from PartVerse mesh annotations (77K training pairs and 588 held-out queries on disjoint object splits), on which 3D-PLOT-LLM reaches caption-to-slots Jaccard 0.459 and Exact-match 13.78%, with a slot-to-caption GPT-4o judge of 44.68. On the 3DCoMPaT-GrIn part-aware grounded description benchmark, 3D-PLOT-LLM outperforms PointLLM, Kestrel, PARIS3D, and SegPoint on every text-output metric, and ShapeLLM on 3 of 4, with up to +3.03 GPT-4o judge over PointLLM. On Objaverse whole-object captioning, adding PartVerse-QA at Stage 2 yields +0.65 SBERT and +1.85 GPT-4o over PointLLM, and tops PointLLM-PiSA on 4 of 5 traditional metrics (SBERT, SimCSE, BLEU-1, METEOR) despite targeting a different (part-grounded) objective. All with under 1M new trainable parameters on a frozen point encoder, an order of magnitude below prior part-aware 3D MLLMs, and no segmentation decoder or bounding-box head.

Concept Bottleneck Models (CBMs) enhance the interpretability of deep learning networks by aligning the features extracted from images with natural concepts. However, existing CBMs are constrained in their ability to generalize beyond a fixed set of predefined classes and the risk of non-concept information leakage, where predictive signals outside the intended concepts are inadvertently exploited. In this paper, we propose Multimodal Concept Bottleneck Model (MM-CBM) to address these issues and extend CBMs into CLIP. MM-CBM utilizes dual Concept Bottleneck Layers (CBLs) to align both the image and text embeddings into interpretable features. This allows us to perform new vision tasks like zero-shot classification or image retrieval in an interpretable way. Compared to existing methods, MM-CBM achieves up to 51.26% accuracy improvement on average across four standard benchmarks. Our method maintains high accuracy, staying within ~5% of black-box performance while offering greater interpretability.

Unlocking the spatial intelligence of multimodal large language model (MLLMs) is crucial for understanding and interacting with the 3D world. Prevailing approaches typically inject spatial priors via external tools, which impose significant inference overhead, or rely on latent feature distillation, which remains uninterpretable and lacks fine-grained geometric constraints. To address these issues, we propose SpatialSV, a framework designed to internalize robust 3D spatial awareness within MLLMs while simultaneously offering inherent interpretability. Deviating from passive feature imitation, SpatialSV employs task-oriented visual supervision, compelling the model to actively lift its 2D visual features into explicit 3D representations, including depth maps, camera poses, and point clouds. Crucially, this 2D-to-3D lifting process provides a transparent window into the model's representations: the resulting 3D reconstructions serve as an intuitive proxy for visualizing and diagnosing the quality of the model's intrinsic spatial knowledge. Extensive experiments across multiple models and benchmarks demonstrate the effectiveness of SpatialSV in enhancing and interpreting MLLMs' spatial intelligence. Furthermore, the framework exhibits strong generalization in semi-supervised settings, validating its potential to leverage unlabeled visual data for scalable, interpretable spatial representation learning.

Multimodal Large Language Models (MLLMs) often lose track of the right image regions during fine-grained spatial reasoning, because a textual query rarely carries any explicit geometric anchor into the pixel domain. Prevailing remedies either rewire the model's weights or pad the prompt with verbose instructions, yet neither reliably pins the language to the correct visual coordinates without eroding the backbone's general competence. We introduce Timage, a paradigm that recasts multimodal understanding as an alignment problem solved at the input: the query is drawn, as a typeset overlay, onto the image itself. The placement and appearance of this overlay are produced by a Constrained Schrödinger Bridge (cSB), an entropic optimal-transport sampler that factorizes layout synthesis into two coupled stochastic stages. The first stage, Region Search, transports noise toward query-aligned image zones while obeying a hard occlusion barrier that protects salient foreground content; the second stage, Appearance Shaping, sizes the glyphs through an ``ink-budget'' regularizer so that the rendered text stays legible and visually balanced. The resulting overlay behaves as an explicit attention beacon that channels the model's focus along spatial semantics. On the VMCBench suite, Timage paired with a modest 7B backbone clearly overtakes far larger proprietary systems as well as parameter-tuned baselines. The study positions deliberate input reconstruction as a powerful, architecture-neutral lever for strengthening multimodal reasoning.

Visual tokens enter Large Language Models (LLMs) as raw, foreign signals. How they are transformed into meaningful representations and interact with the language space depends entirely on the integration architecture. Whether by treating visual tokens as in-context prompts within the input sequence or injecting them directly into the LLM's intermediate layers. A controlled comparison and understanding of how these architectural choices affect visual information and its internal transformation to integrate with the LLM remains underexplored. We provide a fair comparison by evaluating in-context and layer-wise injection VLM integration paradigms under identical training conditions across single image, multi-image, and video benchmarks. In doing so, we uncover a hidden evolution where visual tokens enter the LLM as disguised visual context, raw representations lacking linguistic structure, but are progressively reshaped depending on the integration paradigm, each capturing fundamentally different frequency characteristics of the visual signal. We show that this evolution inside the LLM determines what visual features the VLM can utilize effectively, how visual representations align with the language space, and ultimately how each paradigm performs across different tasks. We further demonstrate that attention allocation alone is insufficient, and that performance is driven by the quality of visual representations at each layer.

Multimodal Large Language Models (MLLMs) have demonstrated remarkable success in various Remote Sensing (RS) tasks. However, their ability to comprehend negation remains underexplored, limiting deployment in real-world applications where models must explicitly identify what is false or absent, e.g., emergency responders need to locate non-flooded routes for evacuation. To comprehensively study this limitation, we introduce RS-Neg, the first benchmark to evaluate negation understanding across region-level to scene-level tasks. Specifically, we design an automated data generation pipeline for RS imagery, using LLMs to synthesize diverse negation queries, and introduce a dynamic visual focus module for verification. Our evaluation reveals that advanced RS MLLMs struggle with negation, exhibiting hallucinations and substantial performance degradation. To close this gap, we propose NeFo, a novel test-time learning method that explicitly incorporates the logical role of negation into the model optimization. Remarkably, using about 5\% unlabeled test samples, NeFo significantly improves the negation understanding of models and shows strong generalization to unseen tasks. Code and data will be released upon acceptance.

Vision-language models (VLMs) often underperform on evidence intensive tasks because decisive visual evidence are small, localized, and easy to overlook, leading to failures in evidence readout even when high-level reasoning is intact. Prior inference-time visual interventions can improve grounding without retraining, but they are largely open-loop and lack a mechanism to verify whether highlighted evidence is actually used. We study answer-span prediction entropy as a model-internal feedback signal and show that naive entropy minimization is ambiguous, since low entropy may arise from evidence-grounded confidence or shortcut collapse. To resolve this ambiguity, we introduce low-entropy anchors and an entropy-shaping objective that reduces answer uncertainty while preserving baseline high-confidence tokens. We instantiate this principle in SPOT-E, a plug-and-play test-time method that produces question-conditioned spotlights, optimized per instance via light-weight tuning based on Group Relative Policy Optimization (GRPO). Across all benchmarks and different VLM families, SPOT-E yields consistent gains and improved robustness under visual corruptions. Code is publicly available at: \url{https://github.com/YinBo0927/SPOT-E}

Vision-language models (VLMs) often generate fluent but visually unsupported descriptions, especially by mentioning objects absent from the image. We propose QK Product Steering, a data-free, training-free, and zero-inference-cost weight edit for reducing object hallucination. The method directly edits the per-head query-key product, the operator that produces pre-softmax attention logits, by suppressing a small number of dominant singular modes in selected middle layers. The edited product is then mapped back to the query weights through a closed-form query-only update while keeping shared key weights fixed, making the edit compatible with grouped-query attention. We further decompose the QK product into symmetric and antisymmetric components to distinguish mutual content-similarity patterns from directional attention patterns. Across three GQA-based VLMs, QK Product Steering achieves an average relative CHAIR$_s$ reduction of $4.0\%$, while matched random-mode controls show negligible change. Interpretability ablations show that the hallucination signal is specific to dominant QK modes and is primarily localized to the symmetric mutual-attention channel. Overall, QK Product Steering offers a simple alternative to decoding-time mitigation, requiring no additional data, fine-tuning, or inference-time overhead while largely preserving general multimodal capability.

World Action Models (WAMs) commonly rely on video generation to bridge visual world modeling and robot control. However, video-based WAMs face three coupled limitations: dense multi-frame future tokens make inference costly, full video prediction spends capacity on action-irrelevant temporal and appearance details, and long-horizon future imagination may introduce errors that mislead action prediction. These issues raise a simple question: Does world action model really need video generation? We propose ImageWAM, a simple WAM framework that repurposes pretrained image editing models for robot action prediction. In contrast to video generation, image editing provides a better-matched prior: it only needs to model a target-frame transformation, focuses on action-relevant current-to-target visual differences, and grounds task instructions to localized visual changes through edit pretraining. In practice, ImageWAM does not decode the target frame at inference time; instead, it conditions a flow-matching action expert on the KV caches produced by image-editing denoising, using them as a compact world-action context. ImageWAM outperforms standard VLA baselines and matching competitive WAMs without additional policy pretraining across different simulator and real-world experiments. It also reduces FLOPs to 1/6 and latency to 1/4 of video-based WAMs. Attention analysis further shows that editing caches focus on task-relevant change regions, supporting image editing as an effective alternative to video-based world-action modeling.

UAV Vision-Language Navigation (UAV-VLN) is typically formulated as a holistic search-and-reach problem, where long-range target discovery and final target approach are optimized and evaluated jointly. This formulation makes it difficult to assess a critical capability of aerial embodied agents, namely whether a UAV can accurately ground a visible target and translate vision-language evidence into precise 3D motion once the target enters its field of view. To address this limitation, we introduce UAV-VLN-FOV, a target-visible navigation task that isolates the see-and-reach stage and enables a more diagnostic evaluation of terminal reaching ability. We further propose 3DG-VLN, a vision-language waypoint prediction framework guided by dynamic 3D direction cues to enhance fine-grained visual grounding and spatial direction alignment for precise target reaching. Specifically, 3DG-VLN adaptively processes high-resolution front-view and downward-view observations to preserve fine-grained visual and geometric details for target grounding. It also updates the target-relative direction online during closed-loop navigation, allowing the agent to maintain spatial alignment with the target and reduce accumulated direction drift. To support this task, we construct a dedicated high-resolution benchmark which contains 2,717 trajectories with target-oriented high-level instructions, high-resolution front-view and downward-view egocentric observations, and continuous 3D waypoint annotations. Experiments show that 3DG-VLN outperforms competitive UAV-VLN baselines, achieving a 13.82\% improvement in success rate. Real-world trials further demonstrate the potential of 3DG-VLN for practical see-and-reach navigation. The source code and benchmark are available at https://github.com/xuefanfu/3DG-VLN.

Memory remains a critical bottleneck for long-horizon robotic manipulation, as standard Vision-Language-Action (VLA) policies often fail when task-relevant cues become occluded or unobservable over time. While existing memory-augmented methods utilize historical context, they either suffer from severe information bottlenecks, incur high latency via decoupled dual systems, or rely on unselective buffers that accumulate massive visual redundancies. To address these limitations, we introduce EventVLA, an end-to-end framework founded on the concept of sparse visual evidence memory that comprises two core components: foundational visual anchors to retain initial and short-term contexts, and a dynamic Keyframe Evidence Memory (KEM) module. Specifically, KEM directly predicts future keyframe probabilities from the VLA's latent embeddings to autonomously capture and store sparse, task-critical visual events. This foresight-driven mechanism empowers the policy to dynamically evaluate the future causal utility of current observations, preserving transient visual evidence before it becomes unobservable. Furthermore, we propose RoboTwin-MeM, a diagnostic benchmark specifically designed to evaluate non-Markovian manipulation tasks with interactive visual evidence. Extensive evaluations show that across 17 memory-requiring simulation tasks and 4 real-world bimanual tasks, EventVLA achieves an average success rate improvement of +40% over state-of-the-art memory-augmented VLAs.

Synthetic data for autonomous driving is surging, powered by diffusion models that promise scalable scene generation. Yet key obstacles remain, as enforcing multi-view and temporal consistency often relies on backbone fine-tuning or added layers, which erodes pre-trained knowledge and weakens text alignment. Models also stay close to the training distribution, struggling under adverse weather and unseen configurations, and fidelity favors frequent over rare classes. We address these gaps with FrozenDrive, a controllable generative framework that preserves a pretrained diffusion models knowledge while achieving strong consistency. FrozenDrive conditions on rich driving-stack signals and text prompts, and introduces knowledge-preserving spatio-temporal attention to impose cross-view alignment and temporal coherence in a single pass within a parameter-free frozen diffusion backbone. An additional object-focused constraint improves per-object fidelity for rare categories. Without any weather- or scene-specific fine-tuning, our model synthesizes globally coherent multi-view driving scenes from text, particularly under adverse and rare conditions, and surpasses prior baselines. On nuScenes, FrozenDrive augmented data significantly improves AD models performance, especially at night and in rain, demonstrating stronger robustness when trained with our scenario-targeted data.

Leveraging Vision Foundation Models (VFMs) for camera-to-LiDAR knowledge distillation offers a promising solution to the scarcity of annotated data needed to represent the immense geometric and kinematic diversity of real-world autonomous driving (AD). However, current approaches typically treat VFMs as black-box teachers, relying exclusively on frame-wise feature similarity. Consequently, they do not fully exploit the teacher's layer-wise semantic structure and global context, as well as the rich spatiotemporal information inherent in LiDAR sequences. We propose HilDA, a self-supervised pretraining framework for LiDAR backbones that better captures the semantic what and geometric where needed for driving tasks. HilDA combines hierarchical distillation comprising multi-layer distillation for progressive semantic alignment and global context distillation for scene-level semantics, with a temporal occupancy diffusion objective promoting spatiotemporal consistency. Models pre-trained with HilDA achieve state-of-the-art results on cross-modal distillation benchmarks and outperform models trained via prior distillation approaches on 3D object detection, scene flow, and semantic occupancy prediction. Code available at: https://maxiuw.github.io/hilda.

Real-world spatial intelligence requires reasoning over a continuous and evolving 3D world, yet existing VLMs and tool-augmented agents largely remain tied to static, stateless inference from isolated visual observations. We introduce \textbf{\textsc{S-Agent}}, a spatial tool-use agentic paradigm for understanding and reasoning over continuous multi-view images and videos. By formulating spatial reasoning as spatio-temporal evidence accumulation rather than isolated frame-level prediction, \textsc{S-Agent} reshapes spatial perception into scene-centric understanding beyond frame-centric recognition. Specifically, \textsc{S-Agent} casts the VLM as a semantic planner that decides what evidence is needed, while a hierarchy of spatial tools and experts grounds objects in 2D, lifts them into 3D geometric evidence, and aggregates this evidence into high-level spatial knowledge (\textit{e.g.}, counting, measurement, orientation, and relative position). Additionally, a temporal memory mechanism, including Scene Memory for maintaining the evolving scene state and Agent Memory for accumulating reasoning context, enables evidence integration across frames and reasoning steps. Comprehensive experiments on multi-view and video spatial reasoning benchmarks show that \textsc{S-Agent} consistently improves both open-source and closed-source VLMs in a training-free manner. Beyond inference-time augmentation, supervised fine-tuning (SFT) on \textsc{S-Agent}-generated spatial trajectories \textsc{S-300K} yields \textsc{S-Agent-8B}, a compact spatial agent that significantly surpasses similar-scale baselines (e.g., Qwen3-VL-8B) and performs comparably to advanced closed-source models (e.g., GPT-5.4 and Gemini 3).

Embodied foundation models are expected to benefit from data scaling like large language models, but face a much tighter data bottleneck. Teleoperated real-robot trajectories remain the dominant pretraining source due to their precise action supervision and embodiment alignment, yet their scalability is limited by high collection cost, acquisition difficulty, and low behavioral and environmental diversity. These limitations have sparked interest in egocentric human video as a scalable, substantially lower-cost, and more diverse alternative for embodied model pretraining. However, its effectiveness compared to teleoperated real-robot data remains underexplored. To address this question, we conduct a systematic study comparing egocentric human video and teleoperated real-robot trajectories as pretraining data sources for embodied foundation models, under fixed post-training and validation protocols. Surprisingly, we find that egocentric data, when processed through a carefully designed filtering and labeling pipeline, is not merely a viable substitute for model pretraining but can lead to superior performance. With the same amount of pretraining data, models pretrained on egocentric data achieve a 24% lower validation loss on real-robot action prediction, as well as 52.5% and 90% higher success rates on in-distribution and out-of-distribution real-robot task execution, respectively. This finding verifies a scalable paradigm for embodied foundation models: pretrain on egocentric human video to learn diverse world representations, then adapt with a small amount of labeled real-robot data for action-space alignment. We hope this study encourages broader exploration of egocentric data and offers guidance for data quality assessment before costly robot data collection.

Composed image retrieval retrieves a target image using a composed query of a reference image and a modified text description. In the fashion domain, this task requires understanding subtle attribute variations such as color, pattern, and texture. However, existing approaches face limitations due to scarce annotated data and simplistic negative sampling. We propose a novel framework that integrates a multi-modal large language model (LLaVA) to generate attribute-aware triplets and introduces a two-stage fine-tuning strategy to enhance contrastive learning. We leverage pretrained vision-language models, such as CLIP-ViT/B32, to generate and concatenate sentence-level prompts with the relative caption and to scale the number of negatives using static representations. Experimental results demonstrate enhanced compositional reasoning and improved fine-grained retrieval behavior, underscoring the feasibility and potential of the proposed framework for fashion retrieval.

Despite significant progress in multi-modal Re-Identification (ReID), existing methods tend to emphasize low-frequency cues. Consequently, they focus on attributes such as color, illumination, and coarse appearance, while overlooking mid and high-frequency structures that encode geometric, textural, and identity-discriminative details. This imbalance leads to incomplete spectral representations and unstable cross-modal alignment. To overcome these limitations, we introduce FUSE, a frequency-domain framework that reformulates multi-modal ReID as a two-stage process of spectral disentanglement and energy alignment. The proposed Spectral Decomposition Module (SDM) adaptively partitions features into low, mid, and high-frequency subspaces, enabling hierarchical spectral modeling. The Cross-Modal Alignment Module (CAM) further enforces energy alignment and subspace complementarity across modalities via frequency-consistency regularization. In addition, FUSE incorporates learnable frequency modulation to enhance robustness under varying illumination and heterogeneous sensor conditions. Extensive experiments on RGBNT201, RGBNT100, and MSVR310 show that FUSE achieves 9.1\% mAP and 9.5\% Rank-1 improvements, establishing an interpretable frequency-domain paradigm for multi-modal representation learning.

While some individuals possess a natural talent for drawing, mastering this skill requires dedicated training and practice. Determining one's skill in the art of drawing requires proper comprehensive assessment. In this paper, we propose a method to measure drawing skill by by matching the hand-drawn image with the original template. Existing techniques often involve complex processes. However, advancements in computer vision allow us to train computers to perform these comparisons at a human-like level, thereby resolving the tedious and overwhelming traditional process. Using computer vision applications, determining image similarity involves identifying the level of similarities in an image with a reference image. We have implemented and analyzed the SIFT feature and Siamese network to measure image similarity. Our results indicate that it is feasible to assess art skill levels. Through feature analysis, we found that SIFT-based key point matching provides a more effective means of detecting drawing skills.

Unlike traditional remote sensing change detection that relies on predefined categories, Open-Vocabulary Change Detection (OVCD) identifies land cover changes flexibly using arbitrary text prompts. However, existing methods suffer from an inherent trade-off when modeling changes: instance-level comparison overlooks fine-grained semantic variations (e.g., partial building extensions), while direct pixel comparison proves unreliable, yielding unstable responses and boundary artifacts due to semantic ambiguity and spatial inconsistency. To this end, we propose an efficient training-free Reliability-Aware Open-Vocabulary Change Detection (ReA-OVCD) framework. It first derives candidate change regions from pixel-wise semantic discrepancies to ensure flexible and detailed localization. To ensure reliability, it subsequently introduces a collaborative refinement strategy to explicitly model change validity from both semantic and spatial perspectives. Specifically, we develop a Semantic Change Reasoning (SCR) module that reassesses changes by jointly analyzing distributional divergence and response variation, enabling the suppression of incidental inconsistencies while preserving reliable semantic shifts. In addition, a Boundary-aware Change Refinement (BCR) module is designed to mitigate artifacts stemming from boundary misalignment and uncertainty through validating whether candidate regions are supported by reliable interior pixels. Extensive experiments across multiple datasets (LEVIR-CD, WHU-CD, DSIFN, and SECOND) demonstrate that our method consistently outperforms state-of-the-art approaches, achieving $\mathrm{F}_{1}^{C}$ improvements of 2.13\% to 9.75\% with higher computational efficiency. The code is publicly available at \https://github.com/Funny0101/ReA-OVCD

We describe our 4th-place entry to the ICRA 2026 GOOSE 2D Fine-Grained Semantic Segmentation Challenge, which reached a composite mean Intersection-over-Union (mIoU) of 69.73% on the official 1,815-image test set. Our model adapts the image encoder of a recent visual foundation model, Segment Anything Model 3 (SAM3), with a lightweight decoder. Beyond this, we contribute two techniques and one empirical finding: (i) a self-distillation scheme that re-uses SAM3 itself, prompted with ground-truth boxes, as a teacher on the classes where it outperforms our own model; (ii) an image-level multi-scale test-time augmentation scheme that restores multi-scale inference for a fixed-input-size model by rescaling the image rather than the model input; and (iii) the finding that an aggressive photometric distortion from a winning 2025 GOOSE 2D entry, transplanted onto our pipeline, is its single largest source of improvement.

Imperfectly supervised video polyp segmentation (VPS) aims to learn dense, temporally consistent masks from inexpensive supervision, including weak annotations (points, scribbles) and semi-supervision with few densely labeled frames. This setting is clinically valuable but challenging due to weak contrast, ambiguous boundaries, motion blur, and specular highlights, compounded by sparse pixel-level guidance. While SAM2 can generate dense masks from sparse inputs, direct pseudo-labeling often yields geometry-degraded masks with boundary leakage, underutilizes temporal consistency, and ignores reliability. To address these issues, we propose ARTEMIS, a unified framework for imperfectly supervised VPS driven by agent-guided reliability-aware temporal mask evolution. ARTEMIS initializes coarse masks from available supervision: SAM2 converts points/scribbles, while dense labels serve as reliable anchors. A debate-and-judge vision-language agent selects reliable temporal anchors under weak supervision, which are propagated bidirectionally with SAM2 to refine unreliable or unlabeled frames. Finally, ARTEMIS trains the segmenter using temporal reliability-aware robust learning, incorporating reliability-guided reference selection, a Reference Prototype Transport Module, and reliability-aware robust loss. These components assess mask reliability, evolve anchors over time, transport target identity across frames, and down-weight noisy supervision instead of discarding difficult samples. Experiments on SUN-SEG and CVC-ClinicDB-612 under scribble, point, and limited-label settings demonstrate that ARTEMIS achieves state-of-the-art performance. Code will be released at https://github.com/wangtong627/ARTEMIS.

Mamba-based models have emerged as a promising alternative for salient object detection (SOD), offering significant advantages in modeling long sequences. However, existing models often fail to explore contextual information and the depth of the entire architecture. This paper introduces U$^2$Mamba, a powerful and innovative U-structured network for salient object detection. We propose multiscale Mamba U-blocks (MMUBs) that enhance the model depth to improve local feature extraction capabilities. Our newly developed nested U-structure, incorporating MMUBs, enables the network to integrate various receptive fields from shallow and deep layers, thereby collecting richer contextual information and longer-range data without being constrained by resolution. Instead of using the traditional deep supervision scheme and top-level supervised training, we propose a hierarchical training supervision method where the loss is computed at each level during the training process. Extensive experiments demonstrate that U$^2$Mamba achieves highly competitive performance against state-of-the-art methods. The source code is available at \url{https://github.com/JL021/U2Mamba}.

Few-shot anomaly detection remains challenging due to limited training data. Multi-modal anomaly detection (MAD) offers a viable solution, leveraging 3D geometric cues to enrich 2D RGB representations and compensate for this scarcity. However, existing MAD methods apply spatially uniform feature processing, conflating stable macroscopic structures with high-frequency localized defect signals, exacerbating cross-modal misalignment and inflating false-positive rates. To overcome this, we present CMDS-AD, a Cross-Modal Dual-Stream Anomaly Detection framework. A LoRA-guided diffusion model generates diverse RGB samples to mitigate extreme data scarcity. For 3D normal augmentation, we employ a pre-trained diffusion model as a normal estimator. Crucially, this estimator inherently acts as a non-linear low-pass filter, directly extracting low-frequency normal representations from RGB inputs. This establishes an auxiliary estimated stream of purely low-frequency information, anchoring robust structural templates and assisting the uncompressed real stream, containing coupled high- and low-frequency components, to precisely isolate micro-defects. A Coordinate-Aware Hierarchical Feature Mapper adaptively aligns cross-modal semantics, while a multiplicative scoring mechanism filters modality-specific noise. Under the extreme 1-shot setting, CMDS-AD achieves absolute performance gains of 5.7% (I-AUROC) and 2.0% (AUPRO) on MVTec 3D-AD, alongside 7.7% and 5.6% improvements on EyeCandies, establishing a new state-of-the-art.

This paper presents Vortex, the multimodal video retrieval system developed by our team, FocusOnFun, for the Ho Chi Minh City AI Challenge 2025, designed to advance intelligent multimedia search and temporal reasoning. The system integrates adaptive keyframe extraction, multimodal metadata generation from vision-language and speech models, and a hybrid retrieval strategy that fuses CLIP and SigLIP2 embeddings through Reciprocal Rank Fusion to balance global and fine-grained semantics. To enhance interactivity, Vortex incorporates Rocchio-based relevance feedback and a multi-stage temporal search mechanism for sequential event alignment. Built on Milvus and Elasticsearch, the architecture enables scalable indexing and efficient retrieval. Evaluated in the official competition, our FocusOnFun team's system achieved a score of 79.6/88 (90.5\%) in the Preliminary Round and was further evaluated in the Final Round, achieving an `Excellent' overall performance with `Outstanding' results in the question-answering (QA) task. This demonstrating the complementary strengths of CLIP and SigLIP2 and confirming the effectiveness of the hybrid retrieval approach. The system establishes a robust foundation for future research in intelligent, context-aware, and interactive video retrieval.

Recent progress in vision-language models has enabled the processing of increasingly long video sequences, but the ability to handle extended token streams does not translate to understanding of narrative structure in long videos. Existing long video benchmarks focus on needle-in-a-haystack retrieval rather than evaluating how low-level actions form events, how events interact across time, and how narratives progress, for example, whether a model can connect an early setback, such as a job loss to a later relationship breakup, despite long gaps, intervening scenes, or flashbacks that reframe what occurred. We introduce NEST (Narrative Event Structures in Time for Long Video Understanding), a dataset of 1005 full-length movies (avg. 98 minutes), each annotated with 102 multimodal narrative events grounded in visual content, dialogue, and audio. NEST captures multimodal narrative events with structured annotations grounded in visual content, dialogue, and audio, and links them through relations that reflect narrative structure, including temporal ordering, hierarchical composition, and long-range dependencies. We introduce baselines for event trigger detection (ETD), event localization (EL), event argument extraction (EAE), and event relation extraction (ERE). The benchmark is highly challenging for grounded event discovery, with ETD below 8%, EL under 6%, and EAE below 11%. In contrast, ERE is more tractable once events are given, reaching 35.45% F1 zero-shot and 44.42% F1 after fine-tuning.

Streaming VideoLLMs must continuously process incoming video while maintaining low query latency, making both video-ingestion throughput and query-time responsiveness critical for real-time deployment. Existing methods largely focus on accelerating individual modules, such as visual encoding, token pruning, or KV-cache compression, but provide limited insight into whether the resulting system can sustain real-time streaming performance. We formulate streaming VideoLLM inference as a coordinated pipeline spanning visual preprocessing, visual encoding, token dropping, and LLM prefilling/decoding. Building on this formulation, we propose ViCoStream (Video Coordinated Streaming), a stage-wise coordinated streaming framework that combines chunk-wise execution, CUDA-stream overlap, visual token control, bounded visual attention, and query-side retrieval to bound per-chunk computation and memory costs. We further provide a systematic study of bottleneck migration, revealing how chunk size, token retention, attention locality, and retrieval scope shape the throughput-accuracy trade-off. Experiments with Qwen2.5-VL-3B/7B-Instruct across multiple streaming benchmarks show that ViCoStream achieves 134 FPS video throughput and less than 50 ms TTFT on a single A100 GPU while maintaining accuracy close to full-history baselines.

In multimodal video reasoning, reinforcement learning-based methods typically rely on simplistic and inflexible reasoning-length control strategies that fail to adapt to the model's evolving competence. This mismatch may suppress necessary exploration at early stages, while encouraging redundant reasoning and inefficient decoding once the model becomes more competent. In this paper, we propose CARE, a competence-aware reward shaping framework for adaptive reasoning length optimization in multimodal reasoning. Specifically, CARE maintains a smoothed competence estimate via an exponential moving average of pass rates, and uses it to route training into progressive stages that shift the reward preference from exploration-oriented long-form reasoning to efficiency-oriented concise reasoning. To avoid conflating verbosity with intrinsic task complexity, CARE further normalizes reasoning effort with batch-level statistics, and introduces a posterior amplifier to strengthen reward signals for unexpectedly strong performance on historically difficult samples. The proposed mechanism is seamlessly integrated into the GRPO training pipeline and incurs no additional inference-time overhead. Extensive experiments on multiple video reasoning and general video understanding benchmarks demonstrate that CARE consistently improves reasoning accuracy, stabilizes reinforcement learning, and significantly enhances token efficiency. Moreover, CARE exhibits a characteristic inverted-U trajectory of reasoning length during training, and yields shorter yet more informative reasoning traces at convergence, indicating effective adaptive allocation of reasoning budget. We provide the source code for our proposed CARE framework and experiments at https://github.com/1Pansy/Video-CARE.

Recent online video instance segmentation (VIS) methods have achieved impressive results, thus becoming the preferred approach to segment instances in videos. Despite the resurgence of impressive single image models, the online (or semi-online) VIS approaches outperform single-image models (e.g., based on SAM) by using long sequences of densely annotated frames during training. However,such a training setup of VIS is expensive in the sense of compute as well as dense annotations required. In order to solve these major flaws, we argue that the effective modeling of the instances and their evolution in videos do not require densely annotated frames. To that end, we propose a simple and effective module, called Past-frames Feature Propagation (PFP) which aggregates low-dimensional features from the image encoder of multiple frames. This simple low-compute module provides tremendous learning capability in using sparse video frame labels for end-to-end training. Combined with a light-weight frame-specific Instance Queries, our Sparse frame Annotation VIS (SA-VIS) significantly improves performance over its baseline. Most interestingly, our simple design that avoids complexities effectively bridges the gap in accuracy between training on sparsely and densely annotated video sequences. This translates to a mere 0.4% drop in performance of SA-VIS when using annotations for only 1/5 of the images in the dataset. Empirically, SA-VIS shows strong improvements over the baseline on YouTube-VIS 2019/2021/2022 and Occluded VIS (OVIS) and an over 1% improvement in AP on the state-of-the-art in a limited annotations scenario.

Pose-flow video anomaly detectors are attractive for one-class surveillance because they provide likelihood-based rankings for tracked skeleton windows. However, a single likelihood score may hide multimodal normal behavior and be sensitive to pose-observation noise. We study a frozen-detector setting in which the pose-flow backbone, cached skeleton tracks, and evaluation pipeline are fixed. Reliability-Aware Prototype Calibration (RPC) is a post-hoc score calibration method for this setting. It adds a standardized nearest-prototype deviation in the frozen latent space to the standardized flow score, and uses keypoint confidence only to gate this added geometric evidence. Thus, RPC preserves the original density signal while correcting the ranking with empirical normal-mode structure under pose reliability. Across two frozen pose-flow backbones and four datasets, RPC improves frame-level AUROC in all eight backbone-dataset pairs, with gains ranging from 0.34 to 4.49 percentage points and averaging 2.03 points. Ablation and reliability analyses show that prototype deviation is the main corrective signal, while reliability gating is most useful when pose observations are less trustworthy. These results suggest that lightweight post-hoc calibration can strengthen cached pose-flow systems when retraining or reproducing the full pose pipeline is impractical.

Egocentric video understanding is inherently limited by the narrow perspective of wearable cameras: a single viewpoint, a single modality, a single model cannot capture the full richness of human action. We argue that a truly expressive egocentric representation must subsume complementary knowledge across viewpoints, modalities, and foundation model representations, yet remain deployable from egocentric video alone. To this end, we introduce a hierarchical multi-teacher distillation framework that produces UNIEGO, a unified egocentric encoder trained with nine teachers spanning ego-exo viewpoints, RGB, depth, and skeleton modalities, and four foundation models. Rather than distilling directly from heterogeneous teachers whose incompatible architectures and feature geometries induce conflicting gradients, our framework interposes a layer of representation-specific Proxy models that translate diverse teacher knowledge into a homogeneous egocentric space. A second distillation stage, Selective Proxy Distillation (SPD), then adaptively selects, for each training sample, the subset of proxies that are both correct and confident, distilling exclusively from reliable supervision and suppressing erroneous signals. SPD is further stabilized by initializing UNIEGO as a learned convex combination of proxy parameters, placing the unified model in a well-conditioned region of the loss landscape before distillation begins. UNIEGO achieves state-of-the-art performance across three egocentric video understanding tasks - action recognition, video retrieval, and action segmentation on three challenging ego-exo benchmarks, outperforming naive multi-teacher distillation baselines and demonstrating that structured, proxy-mediated knowledge transfer yields richer and more discriminative egocentric representations.

Long Video Question Answering (LVQA) requires identifying sparse, query-relevant evidence within hours-long untrimmed videos. Existing approaches either process videos densely with large vision-language models (VLMs), incurring prohibitive computational cost, or rely on sparse caption-based reasoning, which often misses temporally localized and motion-centric evidence. We introduce TimeProVe, a cost-efficient hybrid framework for temporally grounded reasoning in long videos. TimeProVe first employs lightweight modules to generate action-grounded answer--evidence hypotheses and subsequently invokes an expensive VLM only for targeted verification. The core of our framework lies in the Action-based Candidate Evidence (ACE) module, which converts temporally localized actions into query-conditioned candidate answers and supporting evidence windows through lightweight LLM reasoning. We further introduce OpenTSUBench (OTB), an open-ended benchmark designed to evaluate temporally grounded reasoning in real-world Activities of Daily Living (ADL) scenarios. Experiments show that TimeProVe outperforms the strongest baseline on OTB by 7.3%, while reducing VLM calls by 75% and inference cost by 93%. Furthermore, without explicit temporal grounding training, TimeProVe achieves competitive performance on Charades-STA, and reaches state-of-the-art results when enhanced with grounding VLMs.

Precise 3D spatial orchestration in text-to-video generation remains a significant challenge, particularly for multi-object scenes where semantic layout and temporal dynamics are often entangled. While existing depth-conditioned models achieve good structural fidelity, they necessitate dense, frame-accurate guidance that is labor-intensive to author for dynamic events involving deformable objects. We present LooseControlVideo, a framework that enables intuitive and expressive control by using sparse, oriented 3D boxes as a "blocking" proxy. This allows users to author high-level layout and trajectory while leveraging a video generative model to generate realistic occlusions, dynamics and interactions. We achieve this by fine-tuning a Wan 2.2 backbone on a video dataset annotated with DNOCS, a novel encoding for 3D size, orientation and depth-ordered occlusions. Furthermore, our method allows for localized refinement, such as adjusting a jump trajectory or adding an interaction, with minimal disruption to the global scene context. Extensive evaluations on the nuScenes, HO-3D, and BEHAVE benchmarks demonstrate that LooseControlVideo significantly outperforms existing 2D-box and flow-based baselines. Our findings indicate a 1.2x to 3x improvement in Trajectory Error; 2x improvement in Rigid Motion Consistency; and a 1.5x to 2x increase in Occlusion Accuracy over current state-of-the-art layout-conditioned models, demonstrating that oriented 3D primitives provide good geometric prior for complex, multi-agent video authoring.

Multi-representation diffusion models can improve visual synthesis by denoising complementary views of an image, but their performance depends critically on the asynchronous schedule that determines when each representation is denoised. We propose to learn this schedule. Our method formulates asynchronous flow matching over multiple representation spaces and uses a schedule-corrected objective that keeps each representation's local noising-time weights fixed as the schedule changes. We instantiate the schedule with a flexible parametric class that is convex and monotone by construction, and learn it using a fast joint probe with less than 1% additional training compute. On ImageNet 256x256, the learned schedule substantially improves both convergence speed and final quality under a matched 675M-parameter XL backbone. With AutoGuidance, our 200-epoch model reaches FID 1.05, matching the 800-epoch SFD-XL baseline with 4x less training. Training to 600 epochs further improves to FID 1.02, outperforming the 1B-parameter SFD-XXL result of FID 1.04 while using a smaller model. In the unguided setting, our 200-epoch model reaches FID 2.37, already below the best 800-epoch SFD-XL result (2.54) at 4x less training, and improves to FID 2.14 at 600 epochs. Code is available at https://github.com/bsq532087/LWD

Diffusion models have achieved remarkable success in image and video generation and editing. While recent studies have extended these efforts toward motion editing, simultaneously transforming both motion and location-despite its practical importance-remains largely unexplored. To better understand robust motion-location editing, we first analyze the fundamental factors that degrade its quality. Based on this analysis, we propose TeleMorpher, one of the first one-shot frameworks to the best of our knowledge, for simultaneous motion-location editing. Our approach leverages motion priors, a target motion-centric video generated from an off-the-shelf model as motion-editing guidance, and the ground truth motion to enable more controllable and precise motion-location editing. Via this, our framework works as follows: (1) we first disentangle the protagonist and the background via pre-trained segmentation and inpainting models. (2) Then, we introduce a training-free pose warping that edits the protagonist's motion with the motion prior as the guidance. (3) The result of warped motion video is directly injected into a baseline motion editor during inference, mitigating the difference between source and target motions while preserving the appearance of the source video. (4) To enhance the reliability of quantitative evaluations, we propose two new LPIPS-based metrics that measure the background consistency before and after the motion editing and the fidelity of motion editing performance via measuring the difference between the extracted protagonist's skeletons from source and target videos. Experiments with in-the-wild videos and the TaiChi dataset demonstrate that TeleMorpher achieves superior performance across both quantitative and qualitative measurements (real-human evaluation), underscoring its effectiveness.

This paper addresses the challenge of one-shot novel view and pose human image synthesis. The existing methods transfer the reference human image to a target pose using a set of 2D pose keypoints or synthesize human images based on generalizable human NeRF which uses human model priors to extract point-wise features. However, pose transfer based methods can not handle complex human pose using ambiguous 2D pose as the condition, while generalizable human NeRFs may be inaccurate to recover occluded/invisiable human parts without extracted reliable features. To solve these problems, we propose a novel approach for novel view and pose synthesis from a singe human image via conditional denoising diffusion model. Our diffusion model divides the novel view and pose synthesis problem into a sequence of conditional denoising steps. Specifically, to generate humans with complex and arbitrary poses, we introduce 3D human priors, i.e., 3D normal map and color prompt, as geometry and color conditions into the generation process. By transferring the reference human into the target human with a series of diffusion steps, our diffusion model enables high-quality synthesis including the occluded/invisible parts. Further, we propose a self-reconstruction based customized refinement to enhance fine details when tested on novel persons.Experimental results on different public datasets demonstrate that our approach significantly outperforms previous methods and also shows better generalization ability across datasets. The code will be made publicly available at https://github.com/Yankeegsj/3DPGDM.

Scaling robot policy learning for autonomous surgery is challenging, as expert demonstrations are expensive and in vivo exploration poses substantial safety risks. Surgical world models address this by generating realistic, action-conditioned future frames from an initial observation, but existing methods exhibit two persistent failure modes: spatial interaction incoherence, where visible instrument contact fails to induce spatially consistent tissue deformation, and temporal fidelity collapse, where prediction errors compound across autoregressive rollouts and progressively corrupt visual quality. We present SurgVista, a surgical world model that mitigates both failures through two training recipes. Deformation Consistency Regularization extracts scene-point trajectories from training videos and enforces cross-frame coherence through latent contrastive learning, strengthening physically consistent instrument-tissue dynamics. Drift Adaptation Training mitigates long-horizon drift by perturbing conditioning frames with online prediction residuals and photometric augmentations calibrated to long-horizon drift statistics, sustaining visual fidelity over extended rollouts. To enable rigorous evaluation, we further introduce SurgWorld-Bench, featuring diverse procedure types, long-range rollouts, and decoupled metrics for instrument-motion accuracy and tissue-response fidelity. Extensive experiments show that SurgVista consistently outperforms state-of-the-art methods across visual quality, temporal consistency, and interaction fidelity, with gains widening as the prediction horizon grows.

Traditional animation production relies heavily on manual drawing and iterative refinement, particularly for key-pose design, in-betweening, and character coloring. While existing animation and video generation methods have made notable progress, they typically depend on RGB boundary frames, dense frame-wise conditions, or complete sketch sequences, limiting their applicability under low-cost input conditions. We present SketchKeyAnime, a video diffusion framework for generating structurally controllable, appearance-consistent, and temporally coherent animations from sparse key-sketch inputs. Given a single reference RGB image and a few temporally indexed key sketches, SketchKeyAnime introduces a dual-branch conditioning mechanism to encode local geometric constraints alongside semantic-temporal context. It leverages Sketch Cross Attention to fuse reference image and sketch conditions with learnable gating, and incorporates an Adaptive Weighted Loss to strengthen supervision on key-sketch frames and line-art regions. Experimental results on the Aesthetic subset of Sakuga-42M show that our approach consistently outperforms representative animation interpolation and sketch-guided generation baselines. Compared to the best-performing baseline, SketchKeyAnime reduces EDMD by 31.9\% and FVD by 9.5\%, demonstrating superior sketch fidelity and temporal coherence, while achieving the best overall performance across most quantitative metrics. These results validate the proposed framework and highlight its potential for low-cost, highly controllable animation creation.

Most diffusion and flow-matching generators define the prior, probability path, and prediction target in the same representation space. Latent diffusion improves efficiency by moving this path into an autoencoder latent space, but the final sample is still produced by a separately trained decoder. This separation creates a mismatch: the generator is optimized for latent-space prediction, while final quality depends on how the decoder handles generated latents that may differ from clean encoder outputs. We introduce CrossFlow, a cross-space flow formulation that maps noisy latent inputs directly to pixel-space images. The key technical step is a velocity-free one-step objective: the latent trajectory defines the training path, but the supervised prediction is an image rather than a latent displacement. This lets one model act both as a one-step latent-to-pixel generator and as a decoder replacement for latent diffusion pipelines. On class-conditional ImageNet-1k at $256\times256$, CrossFlow-XL achieves 1.62 FID with one function evaluation. Ablations show that the latent encoder and pixel-space perceptual and adversarial losses are important for fidelity. These results indicate that cross-space flow objectives can combine the efficiency of latent representations with direct pixel-space supervision, without requiring a separate decoder at inference.

Latent Diffusion Models (LDMs) have become dominant in visual synthesis, but their quality-compute trade-off is largely constrained by the tokenizer's fixed compression ratio. Variable-length tokenizers (VLTs) promise adaptive compression by varying token counts, allowing diffusion models to flexibly balance quality and compute. However, conventional VLTs modulate length by truncating ordered token sequences, which makes token semantics depend on token position and breaks representational alignment across lengths. This leads to a cross-length shift in the latent distribution that hinders a single variable-length diffusion model from operating effectively. To address this, we propose a novel variable-length tokenizer that modulates length by merging tokens. We show that encouraging similar tokens to merge enables direct cross-length representation alignment when the diffusion transformer operates according to the merging pattern. Since conventional merging methods are data-dependent, making the merging pattern inaccessible during generation, we introduce learnable global merging, which is data-independent, to ensure compatibility with diffusion transformers. On ImageNet 256$\times$256 generation, our merging-based variable-length tokenizer integrated with a diffusion transformer achieves a superior gFID-compute trade-off compared to prior VLT methods. Code is available at [this https URL](https://github.com/movinghoon/lgm)

Video world models are moving toward preserving an observed world under controllable camera and object motion while allowing its environmental state to change. Yet these controls remain isolated, and weather generation typically relies on a source video or reconstructed scene that already specifies future structure. We study a first-frame-anchored source-to-state setting, where the model starts from a single image and follows explicit camera and object controls and an optional weather instruction, then generates a video that either preserves the source world or transfers it to a target weather state. To address these challenges, we first build HoloStateData, a state video dataset that turns diverse videos into unified control samples for camera, object, and weather supervision. Second, we introduce Holo-World, a unified controllable video world model that jointly controls scene from a single image. Its Unified Scene Adapter factorizes world preservation and weather transfer into distinct parameter subspaces, using rendered background, geometry buffers, and object controls to maintain controlled scene structure while modeling weather-dependent appearance and particle effects. Additionally, Scene-Weather Decomposed CFG guides scene and weather residuals separately, strengthening target weather effects without over-amplifying the full condition. Quantitative and qualitative experiments demonstrate that Holo-World maintains precise camera and object control with consistent scene structure while transferring scenes into diverse target weather state, outperforming video-to-video weather editing baselines on weather-state generation. Our project page is available at \url{https://xiangchenyin.github.io/Holo-World/}.

Makeup transfer models enable fun augmented reality (AR) experiences as well as virtual try-on (VTO) for online makeup shopping. While recent state-of-the-art diffusion based solutions such as Stable-Makeup dramatically improve the accuracy and realism of makeup transfer, they still face limitations in identity and skin color preservation, making production-level VTO for makeup shopping unrealistic. In this work, we propose MakeupMirror, a diffusion-based approach to makeup transfer that makes significant progress towards preserving facial features and skin tone. We introduce several technical innovations over Stable-Makeup: (1) integration of facial geometry conditioning with ControlNets to maintain facial fidelity; (2) region-specific makeup transfer control to enable precise makeup application across facial regions such as skin, eyes and lips; (3) skin tone-based makeup transfer modulation that prevent skin tone alteration in cross-subject transfer scenarios; and (4) integration of a Levenberg-Marquardt Langevin sampler to speed up inference while maintaining generation quality. Our experiments on CPM-Real, Makeup Wild, and (herein newly collected, more diverse) MakeupSelfies datasets show that MakeupMirror improves relative facial recognition similarity by +60%, reduces relative skin tone difference by -50% over Stable-Makeup, with a latency of 0.7s, while achieving expert acceptance rate of 94% across core facial identity preservation criteria.

Cinematic compositing aims to integrate green-screen characters into novel environments while maintaining physical and photometric realism. Previous methods often fail to capture the complex bidirectional interactions between characters and their surroundings, which we characterize as Character-to-Environment (C2E) physical interaction and Environment-to-Character (E2C) lighting harmonization. To address this, we propose an end-to-end video diffusion framework that jointly models C2E and E2C interactions, specifically handling the challenges of interactive props. Our approach introduces a tri-mask-guided architecture with RGB-D joint denoising to ensure physically consistent interactions among the character, props, and environment. We further develop an efficient prior-driven data curation pipeline to construct high-quality relighting pairs without expensive rendering. Finally, a reference-conditioned mechanism enables controllable environment synthesis and precise prop replacement. Extensive experiments demonstrate that our framework significantly outperforms existing methods in cinematic-quality dynamic video compositing.

Evaluating video generation with clean, pixel-based reward models disconnects evaluation from the noisy diffusion process and incurs massive VAE decoding costs. In this paper, we challenge this paradigm by asking a fundamental question: Can a powerful video generator inherently discriminate preferences directly from noisy latents? To answer this, we introduce \textbf{PRISM} (\textbf{P}reference \textbf{R}epresentation in \textbf{I}ntermediate \textbf{S}tates of Diffusion \textbf{M}odels). PRISM employs a lightweight Query-based Aggregation head with a frozen video diffusion backbone to decode preference signals from noisy latents. Surprisingly, PRISM not only achieves SOTA preference accuracy but also unlocks strong noise-robustness, which enables early-stage Best-of-$N$ sampling. This allows for filtering suboptimal candidates at the very beginning of denoising, drastically reducing computation while boosting video quality. We also reveal a strong positive correlation between a backbone's generative performance and its inherent evaluative power, enabling self-improving video backbones.

Conditional diffusion and flow models routinely fail to satisfy the very constraints that define their task. For instance, a depth-conditioned model often produces images whose re-extracted depth disagrees with the input, even though the forward operator--the depth predictor defining the constraint--is available during both training and inference. Existing approaches generally fall into two categories: supervised models that treat the conditioning signal as a static cue and ignore alignment information at inference, and guidance-based methods that consult it through hand-tuned linear updates, typically trading fidelity to the condition against the plausibility of the generated sample. We argue that the fundamental gap in both paradigms is that the model is never trained to utilize its own alignment error. We introduce FlowBender, a closed-loop framework that treats this error as a first-class input, training the network to learn a correction policy conditioned on inference-time feedback. At each step, an unguided look-ahead pass estimates the clean signal, a task-specific deviation is computed via the forward operator, and a refinement pass consumes this signal to produce a corrected velocity. We propose several variants of FlowBender, including a gradient-based formulation for differentiable operators and a zero-order variant for non-differentiable settings such as JPEG compression. For efficient sampling, we introduce a prior-step shortcut that enables closed-loop correction at a minimal additional computational cost. Across image-to-image translation, restoration, and 3D mesh texturing, FlowBender consistently outperforms standard supervised baselines, alignment-loss-augmented training, and state-of-the-art inference-time guidance, improving fidelity and plausibility simultaneously rather than trading them against each other. Project page: https://flow-bender.github.io/

Style-content dual-reference generation aims to synthesize an image that preserves the structure and semantics of a content reference while adopting the style of a separate style reference.Despite recent progress, this setting remains challenging because models must balance content fidelity, style alignment, and instruction following avoiding semantic leakage from the style reference.A key bottleneck is the lack of large-scale triplet data with clean content-style separation and broad long-tail style coverage.In this work, we propose FreeStyle, a scalable dual-reference generation framework based on community LoRA mining.We treat community LoRAs as compositional anchors for style and content, and design a rigorous generation and filtering pipeline to construct large-scale Style-Reference and Content-Reference triplets across multiple base models.To address content leakage, we adopt a two-stage curriculum with stage-specific disentanglement mechanisms: an attention-level enrichment constraint that suppresses style-reference leakage in the style-transfer stage, and a frequency-aware RoPE modulation strategy that targets positional-correspondence-based leakage in the harder dual-reference stage.We also introduce a benchmark covering both style-reference and dual-reference generation, with evaluations on style similarity, content preservation, aesthetics, instruction following, and leakage rejection. The benchmark incorporates a style-invariant Content Alignment Score (CAS) and introduces a calibrated VLM-based Rejection Score for evaluating generation reliability and leakage suppression.Extensive experiments show that our model achieves a strong balance among style alignment, content preservation, and leakage suppression.

Autoregressive models excel in visual generation by treating images as 1D sequences of discrete tokens, mirroring language modeling. However, this flattening discards the intrinsic 2D spatial locality of visual signals, creating severe computational bottlenecks during inference. We introduce Spatially Speculative Decoding (SSD), a framework that aligns the predictive objective with the natural geometry of images. Rather than predicting only the immediate next token in a 1D sequence, our model simultaneously predicts the adjacent horizontal token and the token directly below it. By capitalizing on this 2D spatial correlation, spatially speculative decoding overcomes the memory wall in visual inference. Our approach accelerates autoregressive image generation by up to 13.3x while maintaining high fidelity on DPG-Bench and GenEval. Our results suggest that respecting the underlying geometry of vision unlocks massive computational efficiencies, paving the way for real-time, high-resolution autoregressive generative models.

Creating 3D visual illusions, a single 3D mesh that reveals entirely different semantics from various viewing angles, is a fascinating but tough challenge. Existing optimization-based methods are slow and can produce oversaturated colors. In contrast, naive stitching approaches fail to produce geometrically coherent objects. This results in visible unnatural seams and semantic leaks. In this paper, we present a fast and training-free framework for generating text-driven 3D visual illusions. Our approach decouples the generation into two stages. First, we propose a cross-space dual-branch denoising process. This process dynamically decodes 3D latents into voxel space for CLIP-guided orientation alignment and Signed Distance Field (SDF) blending, which ensures seamless geometric fusion. Second, we introduce a view-conditioned texture synthesis module that projects and aggregates view-specific 2D diffusion priors onto the fused geometry. Extensive experiments demonstrate that our method generates highly realistic, dual-semantic 3D illusions in just 3-5 minutes. It significantly outperforms existing methods in geometric integrity, semantic recognizability, and efficiency. Project page: https://siang1105.github.io/JanusMesh.github.io/

Efficiently retrieving specific 3D instances from large-scale scenes via natural language prompts remains a formidable challenge in multimedia analysis. Existing approaches predominantly follow a "scene-level embedding" paradigm, which requires distilling high-dimensional semantic features into every 3D primitive. This strategy suffers from a fundamental architectural bottleneck: memory and computational costs scale linearly with scene complexity, inevitably triggering out-of-memory (OOM) failures in city-scale environments. To address this barrier, we propose QueryGaussian, a training-free framework for expeditious and scalable open-vocabulary 3D instance retrieval. Unlike holistic semantic distillation, QueryGaussian employs an instance-level query mechanism that decouples semantic understanding from geometric representation. Specifically, we leverage pre-trained 2D vision models to interpret user prompts and lift segmentation masks into 3D via a concurrent maximum-weight association strategy, ensuring semantic-visual consistency. To mitigate projection ambiguity, we introduce a temporal fusion module with multi-stage adaptive density clustering. Experimental results demonstrate that QueryGaussian not only matches the accuracy of state-of-the-art methods but also delivers a decisive efficiency leap, reducing GPU memory usage by over 70% and accelerating inference by 180x. Crucially, QueryGaussian enables expeditious instance retrieval on city-scale scenes containing tens of millions of Gaussians using consumer-grade hardware.

Recently, vision-language models (VLMs) have made significant progress in 3D scene understanding, driving advances in applications such as embodied intelligence and robotic vision. However, existing approaches typically either rely directly on explicit 3D inputs (e.g., point clouds or RGB-D sequences), or introduce an additional 3D geometry encoder to derive 3D-aware visual tokens from 2D images. Such designs structurally decouple 3D geometric perception from the rich 2D semantics learned via vision-language pre-training, hindering the development of a unified 3D vision-language representation. In this work, we propose Occ-VLM, a novel framework for 3D scene understanding that operates purely on posed RGB images and employs a single 2D vision encoder. Specifically, Occ-VLM reconstructs 3D scene occupancy as an auxiliary geometric prior, which is utilized to spatially associate foreground 2D tokens with 3D space. These tokens are then decoded by a Large Language Model (LLM) for unified scene understanding. Extensive experiments demonstrate that Occ-VLM achieves both accurate geometric perception and robust vision-language reasoning: it attains state-of-the-art performance on multi-view occupancy prediction, while performing on par with 3D-input VLMs on 3D Visual Question Answering (VQA) and 3D dense captioning benchmarks.

Transferring the camera motion of a reference video to a freshly generated one lets creators reuse cinematic moves. Yet reference and target often live at incompatible scales -- a sweep across a galaxy versus a nudge across a desk -- and naively reusing the recovered trajectory yields either imperceptible or violently exaggerated motion. We trace this to a geometric fact: translation-induced image motion scales as ||T||/Z, so a monocular trajectory is meaningful only up to a depth-scale gauge. We distill this into the Parallax Number Pi = ||Delta T|| / Zbar, a dimensionless, gauge-invariant descriptor of how strongly a camera move is felt, and prove that it -- not the raw trajectory -- is the quantity that scale-faithful transfer must preserve. ParaScale is a plug-and-play module that reads Pi off any reference video and re-realizes it against the target scene's own depth, per frame, leaving rotation untouched. Sitting between pose extraction and pose injection, it requires no retraining and drops into any pose-conditioned generator. We further introduce the Parallax Consistency Error (PCE), a scale-symmetric metric that -- unlike the similarity-aligned TransErr -- exposes scene-scale mismatch. Across scale regimes spanning four orders of magnitude and multiple backbones, ParaScale keeps the realized parallax on the identity line and cuts PCE by more than 3x over uncalibrated transfer with no loss of visual fidelity.

Accurate LiDAR-camera calibration is essential for robust multi-modal perception. Targetless approaches avoid manual setup but remain limited by the scarcity of discriminative cross-modal features. Recent methods address this by reconstructing the scene within a differentiable model, enabling extrinsic optimization through dense photometric supervision. Among these, 3D Gaussian Splatting (3DGS) has been widely adopted as a geometric proxy that bridges LiDAR and camera within a single differentiable framework. However, since 3DGS was originally designed for novel view synthesis, existing methods tend to prioritize rendering quality, causing the proxy geometry to drift from the true LiDAR structure. We propose a framework that preserves the metric geometry of the Gaussian proxy by aggregating multi-view LiDAR observations for dense depth supervision and blocking photometric gradients from updating the Gaussian spatial parameters. We validate our method on public driving datasets, where it consistently outperforms existing targetless methods in calibration accuracy.

We present TriFlow, a new generative approach for producing compact 3D meshes with artist-like triangle topology directly from input geometry conditions such as signed distance fields. Our key insight is to represent mesh topology as a nearest-vertex vector field (NVF) defined over the surface, where each point encodes its association to the nearest triangle vertex in the local barycentric frame. We train a latent flow-matching model to synthesize this field, enabling topology generation conditioned on the input geometry. To extract a coherent mesh, we cluster surface regions using the generated NVF and guide a constrained quadric error metric (QEM) mesh simplification with topology-aware optimization. This yields output meshes that closely match the input geometry while exhibiting structured, artist-like connectivity. Experiments demonstrate that TriFlow achieves stronger generalization and significantly improved topology quality compared to state-of-the-art learning-based approaches, alongside 90% lower Chamfer Distance and an 8x speedup.

Sparse novel view synthesis (NVS) remains challenging due to the ambiguity of recovering 3D geometry from few input views. While NeRF- and Gaussian Splatting (GS)-based methods perform well with dense supervision, they often overfit in sparse settings, producing floating artifacts and inconsistent geometry. Silhouette consistency is commonly used as a regularizer, but it remains insufficient, as silhouette-consistent regions can extend beyond the true object geometry. We introduce VisDom, a learning-free geometric constraint that augments classical carving-based visual hull reconstruction by enforcing a minimum multi-view visibility requirement. Specifically, we define a visible domain as the subset of 3D space observed by at least $K$ views and use it as an additional filtering criterion on top of standard silhouette-based reconstruction. This provides a stronger spatial prior in sparse-view settings. We integrate VisDom into both implicit (NeRF) and explicit (GS) pipelines by restricting volumetric sampling and guiding Gaussian placement during optimization. Experiments on three challenging datasets show consistent improvements in sparse-view NVS, enabling high-quality object-centric reconstruction from as few as four input images. Our method is domain-agnostic, requires only silhouettes, and introduces no learned parameters, making it a simple complement to existing approaches. Applying VisDom on top of GaussianObject further improves performance on Omni3D and MipNeRF360, while matching or surpassing it at 22 $\times$ lower training cost.

Text and 2D-conditioning interfaces provide weak, ambiguous control over spatial transformations in image editing -- particularly under large object motions and camera changes. Prior work has used 3D primitives such as boxes, but only as loose conditioning signals indicating approximate object location rather than specifying the transformation. We instead use 3D boxes as structured specifications: the user provides the input and output boxes of the edit, casting editing as a well-posed geometry problem. This ``thinking in boxes'' interface, where each box face is color-coded to convey 3D orientation, gives precise control over translation, rotation, scaling, and viewpoint changes in real images while preserving scene and object identity, and recovering previously unseen object regions. To ground transformations in scene appearance, we introduce a depth-aligned planar floor as a global reference frame, shaded with depth-aware cues. Conditioned on this structure, an image generator produces consistent results under large transformations. Trained in two stages -- on synthetic multi-object scenes and a small set of real-world videos from Objectron -- the system generalizes to complex, in-the-wild real images. Our method operates directly on real photographs and substantially outperforms recent state-of-the-art methods on large 3D edits.

We introduce the first generative foundation model for chest radiograph synthesis trained from scratch at the billion-parameter scale. Existing radiographic AI models often suffer from poor generalisation across patient subpopulations, institutions, and acquisition settings, resulting in limited real-world clinical utility. Controlled, high-fidelity synthesis of chest radiographs is a promising path toward diversifying clinical datasets and evaluating the robustness of diagnostic models. Therefore, we present the largest specialist generative foundation model for chest radiographs to date, with over 1.3B parameters, trained for 1.6T tokens on a curated, heterogeneous dataset comprising 1.2M radiographs and clinical expert-guided metadata. Our model supports controllable radiograph generation and editing across multiple demographic subgroups, acquisition views, and a dozen pathologies. Moreover, we significantly advance the state of the art in radiograph synthesis fidelity, producing images that are indistinguishable from real radiographs to clinical experts.

Echocardiography (echo) is a widely used imaging modality for assessing cardiac function, with Left Ventricular Filling Pressure (LVFP) serving as a critical physiological marker for conditions such as heart failure. Standard LVFP classification into normal \emph{vs} elevated categories relies on the Doppler-derived $E/e'$ ratio, which is operator-dependent and often unavailable in resource-limited settings, motivating methods that infer LVFP directly from B-mode echo. Existing deep learning approaches achieve high performance but remain largely black-box, limiting clinical interpretability. We propose HypOProto, a hyperbolic, ordinal prototype-based framework for interpretable LVFP classification using a frozen, explainable foundation model backbone. HypOProto arranges prototypes along the physiological $E/e'$ scale, placing borderline cases near the hyperboloid root where small angular differences separate similar cases, while normal and elevated cases occupy outward positions reflecting increasing diagnostic certainty. This hyperbolic geometry encodes clinically meaningful ordinal relationships and improves interpretability. We also introduce a novel Hyperbolic Prototype Angular Separation (HyperPAS) loss, enforcing inter-class prototype separation in hyperbolic space. HypOProto achieves SOTA performance while maintaining transparency, and highlights clinically relevant regions in visualizations. This work represents the first prototype-based framework for LVFP classification in echo. Our code can be found at https://github.com/DeepRCL/HypOProto.

Accurate segmentation of thin, tortuous anatomical structures, such as retinal vessels, cerebral vasculature, and facial wrinkles, remains challenging due to low contrast, frequent discontinuities, and severe class imbalance. Although recent convolutional and Transformer-based models have improved performance, they often yield fragmented predictions and fail to recover fine branches. We propose CSWinUNETR, a general-purpose backbone for 2D and 3D thin-structure segmentation. It employs cross-shaped stripe self-attention to model long-range principal-axis context and incorporates cyclic shifts to enhance information exchange across stripes. To better preserve fine-grained details, we further introduce a detail-enhanced multi-scale self-attention module that aggregates contextual features from multi-resolution representations. In addition, we propose sparse-control dynamic snake convolution, which reconstructs reliable dense curvilinear kernels from sparsely predicted control points to better follow tortuous geometry. Extensive experiments on four benchmarks across ophthalmology, neurovascular imaging, and dermatology demonstrate that CSWinUNETR consistently outperforms state-of-the-art methods without task-specific post-processing or topology-aware losses. The code is available at https://github.com/labhai/CSWinUNETR.

Multi-view imaging, such as mammography and chest radiography, is a standard component of clinical practice. However, medical images are often unregistered and contain view-specific artifacts or irrelevant background cues that can obscure diagnostically relevant findings. Many existing methods directly fuse per-view representations, allowing such irrelevant content to contaminate the fused embedding and reducing robustness under varying view configurations. We propose OTCHA, a confidence-aware latent hub token alignment module based on optimal transport (OT) that refines patch tokens before fusion for multi-view classification. OTCHA introduces a set of learnable latent hub tokens shared across views. For each view, we compute an OT plan between patch tokens and hub tokens that jointly considers feature similarity and geometry, and augment the OT formulation with token-conditional dustbins to enable partial matching and discard irrelevant tokens. The resulting transport plan provides token-wise matching confidence, which gates hub-mediated message passing and weights a novel optimal-transport-based representation alignment loss to stabilize refinement. Experiments on three multi-view medical image datasets demonstrate consistent improvements over competing baselines across diverse anatomies and view configurations. Our code is available at https://github.com/labhai/OTCHA.

Computed Tomography (CT) is essential for diagnosing pediatric craniofacial abnormalities, yet poses radiation risks to developing anatomies. Reconstructing 3D CT from sparse bi-planar X-rays offers a low-dose alternative but is severely ill-posed. Existing methods employ geometry-agnostic feature lifting, naively projecting 2D features into 3D without explicit spatial modeling, causing depth ambiguity and degraded osseous boundaries. We present PSCT-Net, a geometry-aware framework with differentiable back-projection. Differentiable back-projection establishes a spatially faithful volumetric prior, alleviating depth ambiguity. An Attention-Guided Projection (AGP-3D) module then learns non-linear voxel-wise correspondences between 2D regions and 3D locations. A Bidirectional Mamba (BiM-3D) module captures long-range volumetric dependencies with linear complexity. We further curate a private institutional pediatric skull CT cohort, PedSkull-CT, comprising normal and pathological cases for internal evaluation, addressing the gap in adult-centric, trunk-focused datasets.

Endoscopic video analysis is essential for gastrointestinal diagnosis and computer-assisted interventions, but video sequences are routinely degraded by specular reflections, motion artifacts, and missing frames. These transient corruptions can distract clinicians, reduce image interpretability, and disrupt downstream tasks such as 3D reconstruction and navigation. Effective restoration therefore requires methods that exploit temporal continuity rather than treating frames in isolation. We introduce a Gaussian Process Prior Variational Autoencoder (GPVAE) framework for endoscopic video restoration that replaces the standard factorized latent prior with a temporal Gaussian process prior, enabling interpolation of missing frames with uncertainty-aware reconstruction. The framework combines endoscopy-specific encoders, including a convolutional EndoVAE backbone and pretrained Vision Transformer encoders from GastroNet-5M, with two scalable GP approximations: Hierarchical Prior Approximation (HPA) and Sparse Precision Approximation (SPA). Specular reflections are handled using a DUCKNet-based masking pipeline that excludes corrupted pixels from the reconstruction objective. On the C3VDv2 colonoscopy dataset, the best GPVAE variants reduced image reconstruction RMSE by 21.9\% on average, and by up to 26.1\%, relative to matched VAE baselines. Downstream trajectory RMSE was reduced by 12.7\% on average across classical visual odometry and a pretrained PoseNet, at an average increase of 27.3\% in training time per epoch. Finally, the GP posterior provides per-frame uncertainty estimates that reflect temporal support and offer a confidence signal for restored frames.

Multimodal Large Language Models (MLLMs) show great potential in medical tasks, but their elicited confidence often misaligns with actual accuracy, potentially leading to misdiagnosis or overlooking correct advice. This study presents the first comprehensive analysis of the relationship between accuracy and confidence in medical MLLMs. It proposes a novel method that combines Multi-Strategy Fusion-Based Interrogation (MS-FBI) with auxiliary expert LLM assessment, aiming to improve confidence calibration in Medical Visual Question Answering (VQA). Experiments demonstrate that our method reduces the Expected Calibration Error (ECE) by an average of 40\% across three Medical VQA datasets, significantly enhancing MLLMs' reliability. The findings highlight the importance of domain-specific calibration for MLLMs in healthcare, offering a more trustworthy solution for AI-assisted diagnosis.

Whole-slide images (WSIs) are widely used for computational cancer prognosis. However, most existing methods primarily focus on in-domain performance and fail to generalize across clinical centers. This limitation stems from their reliance on pixel-derived representations that are highly susceptible to domain-specific artifacts caused by staining protocols and scanner hardware. We hypothesize that high-level pathology semantics, such as tumor grade and micro-environmental architecture, provide a domain-invariant semantic representation that mirrors the robust diagnostic logic of human pathologists. Therefore, we propose a Semantic-Anchored Evidential Fusion Survival (SAEFS) framework, where SAEFS derives semantic anchors from WSIs via Visual Question Answering (VQA), employs a dual-stream WSI evidence extraction architecture, uses Dirichlet-based Subjective Logic to model uncertainty, and fuses semantic and visual evidence through a cautious conjunction rule to avoid overconfident fusion from correlated sources. Trained exclusively on one source domain and evaluated zero-shot across four unseen domains, SAEFS consistently outperforms state-of-the-art models both in prediction accuracy and reliability, improving the average C-index by 10.2%. Quantitative analyses further show that VQA-derived semantic features exhibit significantly lower cross-center divergence than pixel-derived features, highlighting their robustness for cross-center clinical applications.

Attention-based Multiple Instance Learning aggregators in medical imaging are prone to attention concentration, producing overconfident and unstable predictions. We introduce QG-MIL, a gated transformer aggregator that addresses this through four synergistic architectural components: RMSNorm-based pre-normalization, per-head QK normalization, fine-grained attention output gating, and SwiGLU-style feed-forward modules. Together, these design choices stabilize training and distribute attention more uniformly across instances without auxiliary losses, masking, or multi-stage regularization. We evaluate QG-MIL across six benchmarks spanning whole-slide pathology and cell-level hematology, covering two fundamentally different MIL scales. The best-performing QG-MIL variants outperform leading baselines on all six benchmarks, with an average improvement of +6.1 mean macro F1 points. Attention overlays and attention mass analysis confirm more distributed instance weighting. Ablation studies show that while individual components can match the full model on specific datasets, the QG-MIL design provides the most consistent cross-domain performance and tightest variance when compared to selected baselines. We release a configurable implementation to support reproducibility at: https://github.com/unica-visual-intelligence-lab/QG-MIL

Many dense prediction networks rely on additive feature transformations and model higher-order feature interactions only implicitly. Product units provide an explicit mechanism for multiplicative feature modeling, but their logarithmic--exponential formulation can cause numerical instability, which has limited their use in deep dense prediction networks. In this work, we propose Product-Unit U-Net (PU-UNet), a residual U-Net that integrates stable product-unit residual blocks into rich low-resolution stages for medical image segmentation. The proposed formulation combines smooth positivity mapping with log-domain clipping, enabling stable multiplicative feature learning with negligible computational overhead. On ISIC 2018, Kvasir-SEG, and BUSI, PU-UNet achieves Dice scores of 0.942, 0.959, and up to 0.925, respectively. Compared with a matched Residual U-Net baseline, PU-UNet consistently improves Dice and IoU while keeping parameters, FLOPs, and inference latency nearly unchanged, and reduces the image-level false-positive rate on normal BUSI cases from 0.077 to zero. Ablation studies suggest that the gains are associated with product-unit interactions, are strongest under low-resolution placement, and benefit from the proposed stabilization design. These results suggest that stable product-unit residual learning can be an effective way to enhance U-Net-style segmentation networks with explicit multiplicative interactions.

Image quality control is vital for a wide range of downstream applications. Deep learning-based image quality assessment methods typically train classifiers on dataset-specific quality labels, inheriting two limitations: (1) generalization is tied to the labeling criteria of the training set and (2) these methods cannot provide spatial feedback on where the quality is degraded, lacking explainability. In this work, we propose EFIQA, a framework that requires no quality-related supervision and produces spatial quality maps by design. Rather than learning ``what is degradation" from human-annotated labels, EFIQA learns ``what should be there" by leveraging anatomical priors. For fundus photography, we instantiate this as a two-stage approach, by first training an unsupervised anomaly detector via masked anatomical inpainting to identify regions of missing vasculature, and then distilling this prior knowledge into a shallow adapter mapping features of a frozen foundation model to precise quality maps. External-dataset evaluation demonstrates that this label-free approach with minimal adaptation achieves better performance and explainability compared with supervised methods across benchmarks with different quality criteria, highlighting its potential for real-world applications.

Generating high-resolution 3D CT volumes with fine details remains challenging due to substantial computational demands and optimization difficulties inherent to existing generative models. In this paper, we propose the Pixel-Level Residual Diffusion Transformer (PRDiT), a scalable generative framework that synthesizes high-quality 3D medical volumes directly at voxel-level. PRDiT introduces a two-stage training architecture comprising 1) a local denoiser in the form of an MLP-based blind estimator operating on overlapping 3D patches to separate low-frequency structures efficiently, and 2) a global residual diffusion transformer employing memory-efficient attention to model and refine high-frequency residuals across entire volumes. This coarse-to-fine modeling strategy simplifies optimization, enhances training stability, and effectively preserves subtle structures without the limitations of an autoencoder bottleneck. Extensive experiments conducted on the LIDC-IDRI and RAD-ChestCT datasets demonstrate that PRDiT consistently outperforms state-of-the-art models, such as HA-GAN, 3D LDM and WDM-3D, achieving significantly lower 3D FID, MMD and Wasserstein distance scores.

Head and neck cancers (HNC) represent a significant global health burden, with accurate tumor delineation being essential for effective radiotherapy planning. The complexity of the oropharyngeal anatomy, combined with the heterogeneous appearance of tumors on imaging, makes manual segmentation time-intensive and subject to inter-observer variability. Beyond segmentation, predicting long-term clinical outcomes, such as recurrence-free survival (RFS), and determining human papillomavirus (HPV) status from noninvasive imaging, remain challenging yet clinically valuable goals. The HECKTOR 2025 challenge addresses these needs by establishing a comprehensive benchmark for automated HNC analysis using multimodal PET/CT imaging and electronic health records. Building on previous editions (2020-2022), this challenge features an expanded multi-institutional dataset comprising over 1,100 patients from 10 centers worldwide. Participants were tasked with three complementary objectives: (1) segmenting primary gross tumor volumes (GTVp) and metastatic lymph nodes (GTVn), (2) predicting recurrence-free survival, and (3) classifying HPV status. The challenge attracted 35 registered teams, with 15 final submissions evaluated on a held-out test set. Top-performing algorithms achieved a mean Dice similarity coefficient of 0.75 for segmentation, a concordance index of 0.66 for survival prediction, and a balanced accuracy of 0.56 for HPV classification. This paper presents a comprehensive analysis of the submitted methodologies, evaluates their performance across different lesion characteristics, and discusses their implications for clinical translation in automated oncology workflows and decision support systems.

Camera-trap monitoring in African tropical forests increasingly extends beyond closed-canopy interiors to riverbanks, clearings, and park edges. Among available open tools for African forest camera-trap classification, DeepForestVision is the only one providing a matched offline workflow for both photographs and videos, and previous work showed that it outperformed other available baselines on a comparable benchmark. However, it was designed for closed-canopy, ground-level forest interiors and uses a 35-class prediction space that becomes too coarse when deployments encounter arboreal primates, birds, semi-aquatic taxa, or human-associated confounders such as livestock. We present DeepForestVisionV2, an ecology-driven expansion from 35 to 64 prediction classes (61 animal classes plus human, vehicle, and blank) designed to address three recurrent deployment gradients: vertical stratification, scene openness, and anthropogenic interfaces. DeepForestVisionV2 retains the same offline workflow and is trained on 1,535,010 photographs and 243,354 videos from multi-country African tropical-forest projects. Evaluation combines a cross-country cropped-photo validation set, used to assess robustness across sites and camera-trap settings, with three held-out Uganda video benchmarks spanning the targeted gradients. On the validation set, DeepForestVisionV2 reaches 0.86 accuracy, 0.82 macro-F1, and 0.81 balanced accuracy. On the deployment benchmarks, it preserves or improves baseline accuracy despite its harder classification task, while increasing the number of identified taxa from 22 to 29 in forest-interior videos and from 4 to 9 at riverbanks. In the park-edge use case, it raises accuracy from 0.62 to 0.86 and reduces false alarms from 11 to 0. These results show that DeepForestVisionV2 materially improves field utility while preserving robustness across sites, habitats, and camera-trap settings.

Existing weakly supervised semantic segmentation (WSSS) methods in computational pathology rely on a multi-stage paradigm: class activation map (CAM) generation, offline pseudo-mask refinement, and fully supervised retraining. While established, this decoupled approach presents fundamental limitations. The multi-stage process not only incurs high computational training costs but also suffers from error propagation: local texture biases in shallow CNN layers generate false-positive artifacts that subsequent refinement steps often fail to correct. To address these persistent challenges through a simple yet highly effective approach, we propose the Single-Stage Hierarchical Rectification (SSHR) framework. Rather than passively refining CAMs post-hoc, our method proactively purifies intermediate feature representations during the forward pass. We introduce a Hierarchical Feature Rectification Module (HFRM) that utilizes deep global semantic context to filter out local anomalies in shallow layers. This mechanism generates high-fidelity activation maps directly within a single training loop. Experiments on the LUAD-HistoSeg and BCSS datasets demonstrate that SSHR outperforms state-of-the-art multi-stage methods. Furthermore, SSHR reduces training duration by 2 to 5 times. This efficiency minimizes computational overhead and accelerates clinical translation for large-scale histopathology workflows. The code is available at: https://github.com/trongduc-nguyen/SSHR

Automated skin cancer classification from dermoscopic images remains challenging due to heterogeneous lesion structure, strong intra-class variability, and subtle visual differences between benign and malignant cases. Existing CNN/ViT pipelines typically rely on global or patch-level features and often combine patient metadata via late fusion, which limits spatially grounded multimodal reasoning. We present a novel region-based graph learning framework that explicitly models lesions as graphs of spatially coherent superpixel regions represented as frozen CNN features. To capture fine-grained lesion arrangements, we encode inter-regional geometry as edge attributes and introduce a dedicated metadata context node connected to all regions, providing structured integration of demographic/clinical variables within the same relational space. Node representations are updated using our edge-aware graph transformer followed by attention-driven propagation, and a final graph-level embedding for benign-malignant classification. Experiments on four public benchmarks demonstrate that explicit region-level relational modeling and graph-native multimodal fusion yield consistent gains over the state-of-the-art. Consequently, we establish a new graph-centric perspective in which CNN features are modeled as relational nodes and improved through contextual integration, yielding more expressive and robust classifications.

Reliable localisation of the neonatal face is the first step for several video-camera based non-contact assessments such as pain and distress related facial expression analysis, pain scoring, cardiorespiratory signal extraction and cessation of breathing alerts. However, major challenges persist in neonatal clinical environments. Cluttered backgrounds, illumination changes and poor lighting conditions can reduce the accuracy of face detection models. Clinical interventions, monitoring equipment and, in some cases, medical devices can obstruct the face, making visual assessment difficult. We propose a one-stage YOLOv11m-based model tailored for face detection of infants in neonatal clinical environments. We combined multiple publicly available datasets (VGGFace2, CelebA, FDDB, WIDER FACE) to train and evaluate our proposed model. We then fine-tuned our model on a neonatal research dataset involving 228 videos from 114 recording sessions of 113 independent infants. Before fine-tuning, our model achieved an AP50 of 0.87, surpassing the performance of three state-of-the-art general face detectors. Performance improved further to an AP50 of 0.96 after clinical-domain adaptation. Evaluating face detection performance across different datasets remains a challenge due to the lack of publicly available neonatal datasets. Prioritising the creation of such datasets, while upholding appropriate privacy safeguards and ethical standards in their creation and use, would greatly support further progress in this field.

We study how to train visually grounded vision-language models (VLMs) for radiology without manual spatial annotations. We introduce RefRad2D, a large-scale bilingual (German/English) dataset of 1.2M CT and MR image-text pairs derived from clinical practice, with task-specific VQA and spatial grounding subsets generated automatically via LLM-based curation and automated segmentation. Trained on this data, our model RadGrounder jointly performs report generation, visual question answering, and spatial grounding via bounding-box detection or segmentation. On external VQA benchmarks (Slake, VQA-RAD), RadGrounder achieves competitive results with specialized medical VLMs. Adding our clinical data to the training mixture improves open-ended VQA over fine-tuning on the downstream datasets alone, showing the transferability of our dataset. Crucially, adding grounding supervision does not degrade language quality, enabling spatially verifiable outputs at no cost to VQA performance.

Handwritten Mathematical Expression Generation (HMEG) is challenging due to the complex two-dimensional layouts and long-range structural dependencies of mathematical expressions. Existing methods typically rely on explicit spatial supervision, such as symbol-level bounding boxes, which incurs high annotation costs and limits scalability. In this work, we propose DiffMath, a symbol- and graph-aware latent diffusion framework that leverages the hierarchical structure inherent in LaTeX as a structural prior, eliminating the need for positional supervision. First, we design a Relational Abstract Syntax Tree (RelAST), a generation-oriented representation that distills MathML trees into compact triplet sequences [S, R, D], where each token directly encodes a symbol identity, spatial relation, or nesting depth. Second, we introduce MathVAE, which learns structure-preserving latent representations through symbol-aware and relation-aware perceptual regularization, ensuring that the latent space captures both character semantics and spatial topology. Third, MathDiT performs conditional denoising in this structured latent space, further guided by a global symbol-count prior via Adaptive Layer Normalization (AdaLN) to improve structural coherence. Experiments show that DiffMath produces structurally consistent handwritten expressions, achieves superior performance over existing methods, and improves the accuracy of downstream OCR models through synthetic data augmentation.

We present GB-LSR (Global-Bandwidth Local Spectral Representation), a fixed-grid local spectral representation for continuous image reconstruction. The image domain is partitioned into non-overlapping square patches, each carrying coefficients for a truncated Fourier basis predicted from shared convolutional-encoder features. A single trainable scalar bandwidth is shared globally across all patches and images, and reconstruction at any continuous coordinate is a fixed-size basis contraction whose cost is independent of image size. We study three bandwidth-handling variants: a trainable global scalar (main), a fixed global scalar, and a per-patch bandwidth field. On a standardized native-reconstruction benchmark across Kodak, Set14, and Urban100, the main variant outperforms matched-budget amortized LIIF / LTE / WIRE re-implementations by 2.8-3.6 dB PSNR and 0.11-0.15 LPIPS, while running at roughly one-quarter of the slowest baseline's inference cost. The single global scalar suffices empirically: per-patch adaptive-bandwidth alternatives do not improve over it on either a closed-form locality diagnostic or an end-to-end ablation. In a separate arbitrary-scale super-resolution (ASR) extension, GB-LSR achieves competitive PSNR-Y under a canonical-style SR protocol and runs 1.44x faster than LIIF-RDN and 3.25x faster than LTE-SwinIR at x4; within the same extension, a variant trained and evaluated without 4-corner local-ensemble averaging gives a 1.77x speedup with 35% lower peak memory and negligible PSNR change, while additionally widening the RDN encoder from 64 to 96 channels gives a small positive PSNR shift with a 1.58x speedup and 31% lower peak memory. Native-reconstruction claims are scoped to the matched-budget amortized protocol, and ASR claims are scoped to a separate canonical-style SR protocol.

Linear recurrent unit (LRU), designed with a principled formulation for stable linear recurrence, has demonstrated promising accuracy and robustness on long-range dependency tasks. However, its static parameterization and single-scan method limits its applicability to 2D vision tasks. In this study, we propose a LRU-based restoration network with a semantic modulating unit (SMU) to achieve a harmonious balance between performance and efficiency in single-image super-resolution. The SMU plays three key roles: LRU modulation, spatial categorization, and feature enhancement through learned prototype. Extensive experiments demonstrate that our method quantitatively and qualitatively surpasses recent state-of-the-art methods. Notably, our approach achieves superior performance with computational complexity on par with existing methods. The source code and models are available at https://github.com/MingyuChoi-run/LSM

We propose triangular consistency as a first-principled constraint for optical flow, which is agnostic to network architecture, supervision type, and dataset, and applies to both image-pair and multi-frame settings. This simple but powerful constraint is to compose two flows to induce a third flow and enforce consistency among the three. The composed flows may arise from (i) image pairs, yielding cycle consistency; (ii) multiple video frames, producing longer-range motion through temporal chaining; or (iii) image pairs combined with controlled synthetic transformations, which becomes data augmentation. This triangular consistency introduces negligible computational overhead and requires no additional annotations. Since it is derived directly from the geometry of optical flow, it does not rely on model-specific assumptions and serves as a ``universal'' plug-and-play component for optical flow training. Experiments show consistent improvement across supervised, unsupervised, and transfer learning settings.

Latent diffusion models (LDMs) enable efficient image-to-image translation but discard fine spatial details during compression, degrading downstream perception tasks. We identify two bottlenecks: the autoencoder, which loses spatial information, and the conditioning pathway, which further degrades the source signal through naive downsampling. We propose two lightweight, backbone-agnostic fixes: a Source-Conditioned Autoencoder (SCAE) that injects high-resolution source features into the decoder via skip connections, and a Learnable Guidance Encoder (LGE) that replaces naive downsampling with a learned conditioning signal. Evaluated on RGB-to-SWIR translation for driving scenes with two denoiser backbones (U-Net and DiT), our approach improves detection mAP by up to 2x over the latent diffusion baseline, with up to 3.4x gains on small objects (COCO-small, <32^2 px^2), while achieving state-of-the-art FID. We further show that FID and detection performance are poorly correlated, motivating multi-axis evaluation. Results generalise zero-shot to the public RASMD benchmark. We will publicly release test data with annotations, all checkpoints, and training code.

Occlusion-robust scene recovery remains a major challenge in computational imaging, particularly in natural environments where dense foreground vegetation severely limits visibility. We propose a vision-reasoning-guided light field occlusion removal framework that combines the visibility recovery capability of light field integration (LFI) with the semantic reasoning capacity of vision-language models (VLMs). Multi-view observations are first integrated via LFI to suppress foreground occlusions and produce an initial visibility-enhanced representation. A VLM is then incorporated as a conditional semantic prior to restore degraded structures and recover fine details, guided by the observed measurements. To improve recovery consistency and reduce hallucination artifacts, we introduce a multi-sample fusion strategy that aggregates multiple generated hypotheses into a unified estimate. Experimental results on synthetic and real-world datasets demonstrate state-of-the-art performance, achieving the highest average SSIM across four synthetic light field benchmark scenes (4-Syn) and strong generalization across structured and unstructured acquisition settings. These results highlight the effectiveness of combining physical imaging constraints with vision-language reasoning for robust perception under severe occlusion, with applicability to search-and-rescue and exploratory robotic navigation.

The underwater images are captured within diverse water-medium conditions, leading to complex degradation, including color bias, low contrast, and blur effect. Recently, learning-based methods have demonstrated their potential for underwater image enhancement (UIE). However, most of the previous work focus on the training strategy or network design to make the enhanced result aligned well with the labels in datasets, ignoring that the labels are selected from the enhanced results of previous UIE methods and these pseudo-labels are noisy. Consequently, the performance of their models is not satisfactory to a certain extent. However, collecting the true labels of the underwater images is challenging. In this work, we propose a transfer learning-based UIE that does not require underwater images to have paired noisy or true labels for learning. Instead, the UIE task is first divided into global color correction, haze removal, and background noise suppression following the underwater physics. Then multiple types of prior from other vision tasks are leveraged as cross-domain supervision in each step. In this way, a novel UIE is available via transfer learning, and the physics-aligned UIE decomposition provides theoretical soundness. Qualitative and quantitative experiments demonstrate that our proposal based on physics and priors fusion achieves SOTA performance in the UIE task and effectively boosts downstream vision tasks, significantly outperforming benchmark methods. Project repo: https://github.com/Haru2022/P2-UIE.

We propose Mix-QVLA, a task-evidence-aware mixed-precision PTQ framework for VLA models. Mix-QVLA anchors each quantized variant to the full-precision action-token reference decision and evaluates whether quantization preserves task-relevant evidence across key VLA functional boundaries. It computes normalized gradient-weighted task-evidence maps from boundary activations and compares full-precision and quantized maps using evidence-mass and attribution-distribution distortion, capturing changes in both the strength and allocation of decision-supporting evidence. A soft-bottleneck objective aggregates boundary-level degradation into layer-wise sensitivity scores. Mix-QVLA further models sensitivity throughout task execution, capturing phase-dependent shifts in layer importance rather than assuming a fixed sensitivity profile. The resulting evidence- and time-aware scores guide mixed-precision bit allocation under model-size and BitOps budgets. Extensive evaluations on OpenVLA-style policies show that Mix-QVLA improves the accuracy-efficiency trade-off of low-bit VLA deployment. On LIBERO, Mix-QVLA reduces OpenVLA-OFT memory from 15.4 GB to 4.1 GB, retains 96.3 average success compared with 97.1 for the BF16 model, and achieves a 1.52x inference speedup.

Adversarial camouflage in the physical world remains highly challenging, particularly under UAV reconnaissance where targets undergo continuous geometric changes and extreme illumination variations. Existing methods either optimize 2D digital perturbations that fail to generalize to dynamic viewpoints or produce visually unnatural textures that cannot be deployed in real scenarios. Therefore, we propose an end-to-end framework for adversarial camouflage generation that automatically produces wearable adversarial patterns and maintains stable attack performance in real physical environments with changing viewpoints, poses, and lighting conditions. Our method integrates UV-volume rendering with a diffusion-based texture generator, enabling consistent appearance under varying scales, poses, and lighting conditions. To ensure environmental realism, we propose an illumination color consistency estimator that extracts dominant background attributes and guides a natural texture loss to align the generated UV texture with the surrounding environment. A multi-scale dynamic training strategy further enhances robustness against viewpoint shifts and body deformation. Extensive experiments across multiple mainstream detectors demonstrate that our method achieves strong and stable physical attack performance while maintaining high perceptual naturalness, reducing human detection rates without introducing unnatural artifacts.

Text-to-image (T2I) models generate realistic likenesses of some individuals when prompted with their names, raising privacy concerns. However, distinguishing whether a generated face is memorized or fabricated currently requires ground-truth photos, access to training data, or white-box access to model internals, limiting applicability. We introduce a fully black-box behavioral probe that distinguishes between these regimes while requiring no reference photos or prior knowledge of training data. To benchmark this task, we present the NAMESAKES dataset of over one thousand names and faces of public figures spanning a wide range of fame levels, along with perturbed, less famous names. Experiments on state-of-the-art T2I models show that our probe substantially predicts identity memorization and separates memorized from unrecognized names, with further insights into differences across model families.

Deepfakes targeting a high-profile individual, known as Person-of-Interest (POI), are a threat to modern democracies and societies. Current POI deepfake detection methods still struggle to combine robustness to post-processing, efficiency and interpretability, focal aspects of modern deepfake detectors. In this paper we propose CUPID, a POI video deepfake detector that combines UV texture maps, a facial appearance representation derived from 3D face reconstructions, with the representation learning capabilities of the Masked Autoencoder (MAE). Our method does not require any deepfake videos in its training phase. Moreover, it does not even require to include a specific POI in the training set: the combination of UV texture maps extracted from real video frames and the MAE context-guided reconstruction yields a latent space that captures rich and discriminative facial features also for identities unseen during training. In the testing phase, the embeddings extracted from a query video depicting the POI can be matched against pristine reference videos to assess the video authenticity. Furthermore, operating in the UV space naturally provides an additional layer of interpretability. Specifically, we can extract decoded residual maps that highlight which facial regions of a test video deviate most from the identity representation of the corresponding POI. Experiments on four deepfake datasets show that CUPID outperforms current state of the art on most datasets and achieves the best overall robustness against strong downscaling and compression, providing also substantially faster inference. Our experimental code will be released at https://github.com/polimi-ispl/CUPID.

Training-free detectors of AI-generated images promise generator-agnostic deployment without classifier training, yet their reported numbers are rarely compared under a single controlled protocol. We audit two representative training-free scores -- an autoencoder-reconstruction score (AEROBLADE-style) and a noise-perturbation feature-similarity score (RIGID-style) -- plus a naive feature-kNN control, on a common 1,500-image GenImage-derived benchmark spanning seven generators and JPEG compression at quality 70 and 50. The audit yields three cautionary findings. (i) Implementation details masquerade as method differences: replacing the LPIPS backbone (AlexNet -> VGG-16) changes overall AUROC by +0.085, and switching between resize-to-512 and native-resolution preprocessing flips per-generator conclusions by up to 0.38 AUROC. (ii) Score direction is not a property of the method but of its hyperparameters: the RIGID-style score is inverted (AUROC < 0.5) on SD1.5 and Wukong at noise level sigma=0.05, recovers to >0.5 for every generator at sigma=0.01, and collapses to 0.15 at sigma=0.3. (iii) Dataset format bias inflates robustness claims: without unified re-encoding, AUROC under JPEG-50 exceeds the clean condition for the AlexNet-backbone reconstruction score; after bias correction the residual anomaly localizes to a single generator (BigGAN). The audited scores have complementary per-generator failure sets, but naive z-score fusion does not beat the best single score, indicating that exploiting complementarity requires direction-aware combination.

Vision Foundation Models (VFMs) with Vision Transformer (ViT) backbones, such as DINOv2, have become essential for downstream tasks like object recognition and semantic segmentation. The immense computational requirements of backbones often necessitate distillation into smaller architectures for edge deployment. Feature-based knowledge distillation (KD) often suffers from the teacher-student gap; the student struggles to imitate teacher's complex feature map due to its limited capacity. To mitigate this bottleneck, we propose LEAP: Layer-skipping Efficiency via Adaptive Progression, a training curriculum for ViT feature-based knowledge distillation. By utilizing the teacher's intermediate feature maps as a sequence of progressively more difficult targets, our curriculum allows the student to build a foundational representation before tackling higher-level abstractions. Our results demonstrate that this paradigm significantly accelerates convergence through adaptive difficulty selection across various student model sizes and dataset scales. With our curriculum, the LEAP-distilled ViT-S achieves 90.1% accuracy on ImageNet-100, a +12.24% improvement compared with baseline. On ImageNet-1K, LEAP achieves +3.84% and +7.75% improvement for the instance retrieval task on the Oxford and Paris datasets, respectively. Furthermore, the curriculum enables 25.1% savings in training FLOPs and 21% savings in training time on ImageNet-100 by implementing early-stopping for teacher inference during the initial stages of training. Code is available at https://github.com/KevinZ0217/LEAP

With the recent advent of image generative models, synthetic data are increasingly being used to supplement limited real datasets for training computer vision models. However, not all synthetic datasets improve performance equally, and their effectiveness can only be assessed by training a downstream model, which is computationally expensive and time-consuming. This problem is pronounced in the task of object detection, where the required annotations are much more dense due to bounding boxes. In this paper, we propose a pre-computable metric family, dubbed Conditional-Composition Domain Match (CCDM), which serves as a proxy for the relative utility of candidate synthetic training sets for downstream detection. Experiments on the VisDrone-DET dataset show that the CCDM metric families achieve a Spearman correlation of 1.0 with the downstream performance of YOLOv8, clearly outperforming existing metrics for synthetic image evaluation.

Mamba demonstrates strong efficiency in modeling long visual sequences. However, when token reduction is applied to structurally enhanced Mamba variants, these models exhibit a severe performance collapse. We attribute this degradation to the spatially agnostic nature of existing reduction methods, which violate the two-dimensional structural premise required by the selective scanning mechanism. In this work, we propose STORM, a spatial-aware token reduction framework designed to maintain structural integrity throughout the compression process. STORM reformulates reduction into a structured operation on spatial units, enforcing localized constraints to maintain both grid topology and neighborhood coherence. As a plug-and-play module, STORM equips existing reduction pipelines with explicit spatial awareness without any training. Empirical results demonstrate that STORM achieves state-of-the-art pruning accuracy across diverse vision Mamba backbones under training-free settings. Notably, STORM delivers a substantial accuracy recovery on VMamba, outperforming prior methods by up to 63.3\% in top-1 accuracy. Meanwhile, STORM incurs only a 1.0\% accuracy drop on PlainMamba, achieving performance comparable to ViT.

Current machine learning models commonly require large and well-annotated datasets. However, the annotation process often becomes a bottleneck, with increased complexity leading to higher chances of human errors. Within this context, our goal in this paper is to leverage unsupervised algorithms to improve data annotation efficiency for complex semantic segmentation problems in industrial materials science. Previous research has quantified labeling time and others explored unsupervised methods. However, to the best of our knowledge, this is the first study to quantify how much unsupervised algorithms accelerate the labeling process. We aim to validate the extent to which this laborious process can be accelerated, focusing on semantic segmentation tasks that involve annotating each pixel of high-resolution images, such as the microstructure characterization challenge in materials science. Specifically, we demonstrate that by using unsupervised computer vision algorithms, the time required for the labeling process can be reduced from 170 hours to 37 hours, achieving an approximate reduction of 78\%. The dataset we work with includes large images of dimensions 1280x959 and 960x703, which further increases the complexity of the annotation task. Despite these challenges, we create and share the largest public steel microstructure segmentation dataset to date, available under MIT License with permanent DOI, contributing a fully annotated, high-resolution dataset to the field. Additionally, this is the first work to compare the labeling time from scratch (a common approach in previous studies) to the labeling time when using these unsupervised algorithms as a pre-annotation step. Furthermore, we provide a Deep Learning model trained on this dataset, validated by field experts, and deployed in an industrial setting, serving as an initial benchmark for this public dataset.

Multimodal large language models (MLLMs) are increasingly expected to act on visual information, yet the same scene may require different actions under different task contexts. How reliably can a model turn the same visual evidence into the action required by the current context? To answer this question, we introduce \textsc{ROSE} (\textbf{R}eference-conditioned \textbf{O}ddity and \textbf{S}ymbolic \textbf{E}xecution), a controlled benchmark that holds the visual scene fixed while varying region constraints and required symbolic outputs. Through coupled counting and coordinate-action tasks, \textsc{ROSE} tests whether models can infer an implicit majority reference and act on the resulting fine-grained visual evidence under changing contexts. Across nine recent MLLMs, performance drops by as much as 44.5 percentage points from counting-oriented tasks to region-conditioned action, despite 98.8\% human performance. The gap persists on paired scenes and regions for which the same model returns the correct count, while global-click and matched local controls show that coordinate grounding explains only part of the loss, revealing a distinct, model-dependent bottleneck in turning shared visual evidence into context-specific actions.

Generating large images via deep learning requires patching input data to accommodate hardware memory limitations, then assembling output patches, a process that can introduce stitching artifacts when neighboring patches do not align at borders. While these artifacts are known to affect segmentation tasks, their impact on generative models for style-transfer remains poorly understood. We investigated three stitching approaches and two patch dimensionalities (2D vs 3D) using cycleGAN models trained on cryo-electron microscopy datasets. We evaluated both perceptual quality and performance on downstream mitochondria segmentation. Our key findings reveal that: (1) FID scores fail to detect subtle stitching artifacts that significantly impact downstream segmentation performance, (2) 3D models with artifact-free stitching marginally outperform 2D models on downstream tasks, though the improvement barely justifies the computational cost, and (3) 2D models train more stably due to larger batch sizes. Additionally, we demonstrate that ensembling predictions from three orthogonal directions can improve low-quality volumes but provides no benefit for high-quality outputs. These results demonstrate that maximizing generative model performance on large scientific datasets requires careful consideration and mitigation of stitching artifacts, and that perceptual metrics alone are insufficient for evaluating domain adaptation quality in biomedical imaging.

Recent text-to-image generation models have demonstrated remarkable capabilities in synthesizing highly realistic images from text inputs alone. Although existing benchmarks can evaluate the generation capabilities of various models to some extent, they struggle to comprehensively and accurately measure performance across multiple dimensions, often failing to reveal the inherent deficiencies of models in specific categories. To address these limitations, we propose WeGenBench, a novel benchmark designed for the comprehensive, multi-perspective evaluation of text-to-image generation capabilities. Our benchmark comprises a total of 4,000 test prompts across two primary categories, meticulously balanced between Chinese and English to evaluate bilingual and cross-cultural generation capabilities. Beyond macroscopic scene classification, we annotate each prompt with multi-dimensional tags tailored to the distinct content and challenges of each language, thereby refining the generation tasks into more specific sub-categories. Through a cross-dimensional evaluation mechanism leveraging both scene classifications and multi-dimensional tags, WeGenBench can precisely pinpoint model shortcomings in specific generation categories. Furthermore, to measure generation quality more accurately, we design and validate several novel evaluation metrics by integrating Vision-Language Models (VLMs), which assess model performance on domain-specific tasks from three core aspects. Crucially, our approach yields both the assessment outcomes and the detailed reasoning trajectories, facilitating a rigorous verification of the accuracy and soundness of the evaluation results. Finally, we conduct systematic benchmarking on current state-of-the-art methods and provide an in-depth analysis of the limitations present in existing models.

Continual Test-Time Adaptation (CTTA) aims to maintain model performance under evolving target domains by adapting online without labeled data. However, practical deployments often cannot retain the source dataset due to privacy or licensing constraints, and purely source-free CTTA methods tend to become unstable under long-term distribution shift, suffering from compounding self-training errors and catastrophic forgetting. We introduce DO-ALL (Distill Once, Adapt Life-Long), a plug-and-play framework that revisits source information in a compact and privacy-conscious form via Dataset Distillation (DD). Before deployment, DO-ALL performs DD to produce a small set of synthetic distilled anchors that summarize the source distribution. During adaptation, each target sample is matched with its most semantically aligned anchor, which provides a stable reference for various CTTA via source replay, representation alignment, and manifold-smoothing regularization. DO-ALL can be seamlessly integrated into existing CTTA algorithms, consistently improving long-term robustness across CIFAR100-C, ImageNet-C, and the CCC benchmark. This demonstrates the potential of leveraging DD to enable stable and continuous adaptation without retaining raw source data. The code is available at https://github.com/blue-531/DOALL.

Generative artificial intelligence has the potential to improve productivity and transform the production of creative content. However, existing research indicates that image generation models are significantly influenced by biases. This work investigates the inherent biases and language-induced biases present in text-to-image models within the context of occupation-related image generation, complementing established metrics with human preference feedback. We present a comprehensive evaluation of five current text-to-image models: Midjourney v6.1, Stable Diffusion 3 Medium, DALL-E 3, Playground v2.5, and FLUX.1-dev , focusing on gender and ethnicity bias, image quality, and prompt alignment. To facilitate this evaluation, we developed the "Battle-Arena for Fair Image Synthesis" (BAFIS), a platform designed to collect human feedback on bias in generated images. Furthermore, we created a dataset comprising 21,140 synthetic images generated using multilingual prompts, which serves as a basis for our analysis. We further place our results within a broader social context by comparing them to official statistics from the German Federal Employment Agency. Our findings reveal systematic biases in text-to-image models, with established evaluation metrics in partial correlation with subjective user ratings. Thus, our research emphasizes the need for including human preferences to develop fairer and more inclusive text-to-image models.

Federated Learning (FL) in surgical video AI enables collaborative model training without sharing sensitive data. However, standard evaluation practices - selecting the "best" global model based only on validation data from participating hospitals - can lead to suboptimal deployment choices. We identify this critical failure mode as performance leakage, where the selected model overfits internal federation data and fails to generalize to unseen institutions. We propose GEN-Guard, a practical post-hoc framework to detect and correct generalization failures in federated surgical AI. It integrates Generalization Detection via Client-Blocked Evaluation (CBE), which validates performance on isolated client distributions to prevent performance leakage, and Generalization Correction through Disagreement-Aware Distillation (DAD), which learns adaptive feature-level corrections for cross-institutional robustness. Both components operate after standard FL convergence while providing robust support for zero-shot adaptation to unseen environments. We first quantify the severity of performance leakage, observing Model Selection Failures (MSFs) exceeding 80% under standard evaluation. GEN-Guard is evaluated on two multi-center clinical challenges: surgical phase recognition in laparoscopic cholecystectomy and polyp segmentation in colonoscopy. Across both datasets, GEN-Guard consistently corrects these failures, improving in-federation F1 scores by up to 2 points, unseen-institution performance by up to 3 points, and worst-case institutional performance by 3-9 points. Performance leakage represents a systematic and previously under-recognized risk in federated surgical AI. GEN-Guard provides a practical solution for detecting and correcting such failures. By improving cross-institutional robustness and zero-shot generalization, it strengthens the reliability of FL for real-world surgical deployment.

Building footprint extraction is a fundamental task in photogrammetry, remote sensing, and computer vision. Recent image-based methods have achieved remarkable progress in extracting vectorized footprints from high-resolution optical imagery. However, optical imagery inherently susceptible to occlusions, perspective distortions, and residual relief displacement, yielding incomplete or misaligned footprint extraction. Furthermore, the lack of explicit elevation information limits its direct applicability to Level of Detail building modeling. In this paper, we present PCFootprint, the first large-scale public dataset for footprint extraction from airborne laser scanning point clouds. PCFootprint comprises \num{33000} tiles derived from the Estonian Land and Spatial Development Board, covering diverse urban and rural landscapes. Each tile spans \qtyproduct{128 x 128}{\m} with systematically aligned vectorized footprints aligned to point clouds. The dataset includes a \num{3000} tiles cross-domain test set for evaluating generalization across geographic regions. We establish comprehensive benchmarks by evaluating mainstream methods. Experimental results reveal significant challenges including high intra-class variance, data imbalance, and noise across complex geospatial environments. We believe PCFootprint will advance future research in building modeling, urban scene understanding, and geospatial analysis. The PCFootprint dataset is publicly available at \url{https://huggingface.co/datasets/Haoyuan-Shen/PCFootprint}.

Multimodal foundation models have advanced rapidly thanks to large optical benchmarks, but comparable resources for synthetic aperture radar (SAR) remain limited. Existing SAR--optical datasets largely rely on low-resolution, intensity-only Ground Range Detected~(GRD) products and do not preserve complex-valued SAR measurements or native acquisition geometry, which restricts physically grounded multimodal learning. In particular, large-scale public datasets combining very-high-resolution (VHR) SAR SLC, aligned optical imagery, and natural-language descriptions are still lacking. We present a VHR SAR--optical--text dataset built from open-access Umbra spotlight acquisitions distributed as Sensor Independent Complex Data (SICD). From around 2,500 worldwide scenes (VV/HH, 20cm--2m native resolution), we standardize all SAR data to an 80cm slant-range grid via band-limited FFT resampling and tile the imagery into 1024 by 1024 patches. For each SAR patch, we retrieve a high-resolution optical tile and warp it into the SAR grid using local coordinate correspondences for local pixel-level alignment. We further generate three caption variants (SHORT/MID/LONG) per sample to support vision--language training and evaluation. Our dataset contains 119,566 triplets (complex and amplitude slant-range SAR patch, aligned optical patch, natural-language description) covering 257 locations across 72 countries and a broad range of land types and infrastructures. We release fixed train/validation/test splits and the full preprocessing and baseline code to enable reproducible benchmarks for multimodal alignment on cross-modal retrieval and conditional generation in native SAR geometry. The dataset is publicly available on the Hugging Face Hub at https://huggingface.co/datasets/ONERA/SARLO-80.

The Frechet Inception Distance (FID) is the de facto arbiter of image generation, yet most papers report just a single number from a single trained model using a single sampling seed. How reproducible is that number if we retrain the model, or merely resample from it? In this paper, we treat FID as a random variable on a two-axis panel of training and generation seeds, and measure its variance directly on several hundred SiT networks trained on class-conditional ImageNet 256x256. We report surprising findings: (a) Retraining the model using the same recipe with a different seed moves FID 3.2x more (in Inception feature space) than redrawing samples from a fixed network. (b) That gap is driven by three factors: random initialisation, data ordering, and the per-step Gaussian noise of the flow-matching loss. (c) Increasing compute or model size barely tightens the spread, holding the FID coefficient of variation (CoV) inside a 1-2% band. (d) Per-cell classifier-free-guidance tuning halves the spread but reshuffles which seeds work best, and a lucky training seed reaches the same FID with up to 2x less compute than an unlucky one. Based on these findings, we recommend a new FID evaluation protocol: evaluate under per-cell optimal guidance, treat any FID gap below the empirically measured ~1.3% CoV as inconclusive, and report an error bar over several training seeds rather than a single FID number.

The Caltech Tennis Dataset (CalTennis) is a large-scale video benchmark for evaluating monocular-to-3D pose estimation in the wild. CalTennis comprises over 11 million frames (51 hours) of tennis practice and match play from 40 players, captured with 2-6 synchronized cameras at 60 Hz. It is 10 times larger than existing in-the-wild human motion video datasets and 3 times larger than existing MOCAP-ground-truthed datasets, and it is the first large-scale benchmark to provide synchronized multi-view recordings of expert athletic motion. The multi-view setup enables inexpensive, label-free evaluation of monocular-to-3D pose estimation algorithms. We describe a simple, standardized protocol that enables data collection without specialized equipment or expertise, along with fully automated video calibration and synchronization. Benchmarking state-of-the-art monocular-to-3D pose methods on CalTennis, we find that while 3D joint angle recovery is now quite accurate, all models struggle to estimate depth and foot contact consistently. We further propose two novel performance metrics, footwork and stability, as well as qualitatively study body shape inconsistency. These metrics expose previously underexplored failure modes and point to concrete opportunities for improvement in pose estimation and action analysis.

World models are increasingly regarded as a decisive step toward artificial general intelligence, yet modeling the physical world demands more than rendering convincing frames on demand: it requires an internal world state that keeps evolving over time, decoupled from observation, so that objects endure and events run to their conclusions whether or not a camera is watching, much as the moon holds to its orbit when no one is looking. This requirement is a blind spot of existing benchmarks, which reward surface properties such as fidelity, motion, and camera controllability while never asking whether a generated world keeps evolving once it is unobserved. We introduce \textbf{WRBench}, the first systematic diagnostic benchmark that treats camera motion as an intervention on observability and resolves evaluation into a human-calibrated chain that asks whether the camera executes the requested interaction, whether the scene stays continuous and identifiable while in view, and whether a returning target remains consistent with the event that was set in motion. Across 9{,}600 videos from 23 models spanning four control paradigms, one finding proves stubborn: current systems maintain the observed world as a tracking shot, resuming a returning target in the state at which it was abandoned rather than advancing the event while it went unseen. Because this failure recurs across control paradigms, model families, and increments of scale, robust world-state evolution does not follow from cleaner imagery, tighter control, richer geometric priors, or sheer parameter count We therefore argue that the stability of the physical state kernel and the consistency of worldlines under viewpoint intervention should become first-class objectives of world-model design, so that a world model captures how the world will unfold rather than how the next frame appears.

Event cameras capture dynamic scenes with exceptional temporal fidelity by representing them as a continuous stream of microsecond resolution \textit{events}. Each individual event, however, only carries minimal semantic value, merely signaling a localized brightness change. To derive meaningful signals, downstream algorithms need to quickly integrate cues from a potentially massive torrent of low-information events. Current architectures, however, are easily overwhelmed, struggling to balance capturing fine-grained temporal dynamics and maintaining a manageable data throughput. This paper proposes a framework to re-tokenize event streams into a small set of highly informative \textit{neural events}, each representing a local spatio-temporal context window with a discrete learnable code. Every time this code flips, a neural event is triggered, yielding a highly compressed data stream. We demonstrate that, across object detection and classification, networks trained on neural events are on par or surpass the performance of state-of-the-art approaches while reducing the event rate by a factor of 2.0.