Mitigating Simplicity Bias in OOD Detection through Object Co-occurrence Analysis
通过对象共现分析缓解OOD检测中的简单性偏差
机构 * The University of Hong Kong(香港大学) ; Shenzhen University(深圳大学) ; Shenzhen Loop Area Institute(深圳河套学院)
AI总结 提出基于对象共现的OOD检测框架,通过解耦表示和分治策略区分近OOD,缓解简单性偏差,在多种设置下取得竞争结果。
Comments This paper has been accepted by CVPR2026