[Paper note] MARS: A Video Benchmark for Large-Scale Person Re-identification

最新推荐文章于 2021-08-25 21:21:57 发布

原创

最新推荐文章于 2021-08-25 21:21:57 发布 · 2.5k 阅读

3 ·

CC 4.0 BY-SA版权

文章标签：

#dataset #深度学习 #视频 #行人识别 #re-id

本文介绍了MARS视频行人再识别基准，探讨了其特性、评估方法和实验结果。MARS包含1261个身份，约20,000个视频序列，采用CNN结合度量学习的方法在多个数据集上进行了训练和测试，显示了预训练和度量学习对性能的显著提升。" 52759891,5597018,使用动态规划解决UVa1631 Locker问题,"['动态规划', '字符串算法', 'UVa在线判题', '编程竞赛']

paper
dataset
code for evaluation in Matlab
Authors: Zheng, Liang and Bie, Zhi and Sun, Yifan and Wang, Jingdong and Su, Chi and Wang, Shengjin and Tian, Qi
Su, Chi and Liang, Zhang have worked on re-identification problem quite a time.

Properties

Tracklets: automatically generated by DPM pedestrain detector and GMMCP tracker.
One identity in probe has multiple (averagely about 10 for test set) GT in gallery, mAP is more reaso