arXivDaily arXiv每日学术速递 周一至周五更新

高校专区

Zhejiang University(浙江大学)

2026-08-21 至 2026-08-21 共收录 5
2608.20335 2026-08-21 cs.CV 新提交

4DAnyone: Create Anyone in 4D from a Casual Monocular Video

4DAnyone:从随意的单目视频中创建4D虚拟人物

Yudong Jin, Tao Xie, Qihang Zhang, Zehong Shen, Zhen Xu, Yujun Shen, Hujun Bao, Xiaowei Zhou, Yinghao Xu

机构 * State Key Lab of CAD&CG, Zhejiang University(浙江大学CAD&CG国家重点实验室) Robbyant Chinese University of Hong Kong(香港中文大学) Ant Group(蚂蚁集团) Hong Kong University of Science and Technology(香港科技大学)

AI总结 4DAnyone是一种从单目视频重建4D人体的框架,通过RCP和TCR解决多视图一致性问题,在相关数据集上优于现有方法且泛化性好。

Comments Project page: this https URL (https://4danyone.github.io)

详情

展开后加载摘要…

URL PDF HTML 收藏
2608.20011 2026-08-21 cs.AI cs.CV 新提交

Manifold Drift in Flow Preference Optimization: A Root Cause of Reward Hacking

流偏好优化中的流形漂移:奖励黑客的根本原因

Yansen Han, Shengyi Liao, Yuanxing Zhang, Pengfei Wan, Tao Lin

机构 * Zhejiang University(浙江大学) Kuaishou Technology(快手科技) Westlake University(西湖大学)

AI总结 本文针对流偏好优化中流形漂移导致奖励黑客的问题,提出ThermoDPO及其加权变体,在玩具基准和SD3.5-M数据集上均取得优于FlowDPO等方法的性能。

详情

展开后加载摘要…

URL PDF HTML 收藏
2608.19794 2026-08-21 cs.AI cs.RO 新提交

Towards general embodied intelligence: integrating large language models, knowledge bases, and reasoning capabilities to build the next generation of AI agents

迈向通用具身智能:整合大语言模型、知识库与推理能力以构建下一代AI智能体

Fujiang Yuan, Xia Huang, Lusheng Wang, Jun Ding, Zhen Tian, Yuxin Wang, Shaojie Gu, Yuki Funabora, Yanhong Peng, Zebing Mao

机构 * College of Mechanical Engineering, Chongqing University of Technology(重庆理工大学机械工程学院) James Watt School of Engineering, University of Glasgow(格拉斯哥大学詹姆斯·瓦特工程学院) School of Energy and Power, Jiangsu University of Science and Technology(江苏科技大学能源与动力学院) Magnesium Research Center, Kumamoto University(熊本大学镁研究中心) Department of Information and Communication Engineering, Nagoya University(名古屋大学信息与通信工程系) State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University(浙江大学流体动力与机电系统国家重点实验室)

AI总结 本文综述以LLM为核心的智能系统演进,提出整合LLMs、KBs、RA与具身性的概念框架,明确高效LLM部署等五大挑战,为开发复杂动态环境下的自适应多模态智能体提供路线图。

详情

展开后加载摘要…

URL PDF HTML 收藏
2608.19567 2026-08-21 cs.CV 新提交

Block3D: Efficient Text-to-3D Generation via Block-Wise Diffusion

Block3D:基于分块扩散的高效文本到3D生成

Bowen Cui, Weijie Wang, Zeyu Zhang, Yefei He, Mingda Lin, Haoyu Zhao, Yuanyu He, Donny Y. Chen, Feng Chen, Bohan Zhuang

机构 * ZipLab, Zhejiang University(浙江大学ZipLab) University of California, Berkeley(加州大学伯克利分校) Wuhan University(武汉大学) Monash University(莫纳什大学) University of Adelaide(阿德莱德大学)

AI总结 针对文本到3D生成成本高的问题,提出Block3D分块扩散框架,通过分块生成、联合去噪及置信度引导的块内修正,在保持几何保真度的同时实现5.15倍的速度提升。

Comments Project page: this https URL (https://alexandertsui.github.io/block3d/)

详情

展开后加载摘要…

URL PDF HTML 收藏
2608.19297 2026-08-21 cs.LG 新提交

Holtercare-Bench: A Multimodal Benchmark for Evaluating Long-Term Dynamic ECG Analysis

Holtercare-Bench:用于评估长期动态心电图分析的多模态基准

Yihan Xie, Hanwen Cui, Runze Ye, Juekai Lin, Haoyang Wang, Jinhao Mao, Bo Zhang, Wenqiao Zhang, Xiaogang Guo, Jun Xiao, Lei Zhang

机构 * Zhejiang University(浙江大学) Beijing Institute of Technology(北京理工大学) University of Electronic Science and Technology of China(电子科技大学)

详情

展开后加载摘要…

URL PDF HTML 收藏