Dual representation modeling and progressive contrastive learning for unsupervised video person re-identification

Cong Zhang, Yanzhao Su, Nian Wang, Yunwei Lan, Tao Wang, Aihua Li

Abstract: Highlights•We propose a DRM module to improve the accuracy of tracklet representation.•We propose a DPCL module to preserve the intrinsic variety in tracklet features.•We introduce a MPL module to gradually promote feature embedding.