HieDG: A Hierarchical Discrete Geometry-Guided Framework for Multi-Animal Tracking
HieDG: 一种用于多动物跟踪的层次化离散几何引导框架
机构 * State Key Laboratory of Brain Cognition and Brain-inspired Intelligence Technology, Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所脑认知与脑启发智能技术国家重点实验室) ; School of Artificial Intelligence, University of Chinese Academy of Sciences(中国科学院大学人工智能学院) ; School of Technology, Beijing Forestry University(北京林业大学工学院) ; School of Advanced Interdisciplinary Sciences, University of Chinese Academy of Sciences(中国科学院大学前沿交叉科学学院) ; College of Computer Science, Sichuan University(四川大学计算机学院) ; C²DL, Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所C²DL) ; School of Automation, Chongqing University(重庆大学自动化学院) ; Faculty of Life and Health Sciences, Shenzhen University of Advanced Technology(深圳理工大学生命健康学院) ; School of Future Technology, University of Chinese Academy of Sciences(中国科学院大学未来技术学院)
AI总结 针对多动物跟踪中外观相似、密度高和运动不规则导致的身份关联困难,提出HieDG框架,通过两级残差码本将连续几何信号离散化为结构化令牌,与视觉嵌入对齐并集成到查询中,显著提升身份一致性。
Comments Accepted to ECCV 2026