您好,有最新一些新的TBD的多目标跟踪方法吗?有打算集成一下吗
Zhengzhiyang0000 opened this issue · comments
您好, 最近读论文比较少, 您可以补充一些吗, 我看一下:)
1、MotionTrack: Learning Robust Short-term and Long-term Motions for Multi-Object Tracking(https://openaccess.thecvf.com/content/CVPR2023/papers/Qin_MotionTrack_Learning_Robust_Short-Term_and_Long-Term_Motions_for_Multi-Object_Tracking_CVPR_2023_paper.pdf)这个好像是基于ByteTrack的改进,但是还没开源
2、MOTRv2: Bootstrapping End-to-End Multi-Object Tracking by Pretrained
Object Detectors(https://openaccess.thecvf.com/content/CVPR2023/papers/Zhang_MOTRv2_Bootstrapping_End-to-End_Multi-Object_Tracking_by_Pretrained_Object_Detectors_CVPR_2023_paper.pdf)
3、UTM: A Unified Multiple Object Tracking Model with Identity-Aware Feature Enhancement(https://openaccess.thecvf.com/content/CVPR2023/papers/You_UTM_A_Unified_Multiple_Object_Tracking_Model_With_Identity-Aware_Feature_CVPR_2023_paper.pdf)这个好像是JDE范式的
4、Focus On Details: Online Multi-object Tracking with Diverse Fine-grained Representation(https://openaccess.thecvf.com/content/CVPR2023/papers/Ren_Focus_on_Details_Online_Multi-Object_Tracking_With_Diverse_Fine-Grained_Representation_CVPR_2023_paper.pdf)
谢谢,这四篇文章我都读过,我有空的时候再读一下,看看有没有可能集成