论文标题:Aggregate Tracklet Appearance Features for Multi-Object Tracking论文来源:SPL 2019论文链接:https://www.sci-hub.shop/10.1109/lsp.2019.2940922
如果说 DeepCC 关注的是 ReID 任务在 MTMC 任务中的训练策略设计,那 NOTA 就是针对 ReID 任务在 MOT 任务中的网络框架设计。熟悉 MOT 任务的人应该知道,由于不同质量观测信息和遮挡等问题的影响,我们直接根据给定行人框提取的行人特征并不一定可靠,例如下图中,一个行人框中可能存在多个行人和大量背景信息。
[1] Ristani E, Tomasi C. Features for multi-target multi-camera tracking and re-identification[C]. in: Proceedings of the IEEE conference on computer vision and pattern recognition. 2018. 6036-6046.[2] Chen L, Ai H, Chen R, et al. Aggregate Tracklet Appearance Features for Multi-Object Tracking[J]. IEEE Signal Processing Letters, 2019, 26(11): 1613-1617.[3] Yunzhong H, Liang Z, Zhongdao W, et al. Locality Aware Appearance Metric for Multi-Target Multi-Camera Tracking[J]. arXiv preprint arXiv:1911.12037, 2019.[4] Bergmann P, Meinhardt T, Leal-Taixe L. Tracking without bells and whistles[C]. in: Proceedings of the IEEE International Conference on Computer Vision. 2019. 941-951.[5] Wojke N, Bewley A, Paulus D. Simple online and realtime tracking with a deep association metric[C]. in: 2017 IEEE international conference on image processing (ICIP). IEEE, 2017. 3645-3649.[6] Xu J, Cao Y, Zhang Z, et al. Spatial-temporal relation networks for multi-object tracking[C]. in: Proceedings of the IEEE International Conference on Computer Vision. 2019. 3988-3998.[7] Wang Z, Zheng L, Liu Y, et al. Towards Real-Time Multi-Object Tracking[J]. arXiv preprint arXiv:1909.12605, 2019.[8] http://taodataset.org/[9] https://www.zhihu.com/question/68584669/answer/326110383[10] https://zhuanlan.zhihu.com/p/96999382[11] https://zhuanlan.zhihu.com/p/125395219[12] https://zhuanlan.zhihu.com/p/126558285