Energy clustering for unsupervised person re-identification

Abstract

Due to the high cost of data annotation in supervised learning for person re-identification (Re-ID) methods, unsupervised learning becomes more attractive in the real world. The Bottom-up Clustering (BUC) approach based on hierarchical clustering serves as one promising unsupervised clustering method. One key factor of BUC is the distance measurement strategy. Ideally, the distance measurement should consider both inter-cluster and intra-cluster distance of all samples. However, BUC uses the minimum distance, only considers a pair of the nearest sample between two clusters and ignores the diversity of other samples in clusters. To solve this problem, we propose to use the energy distance to evaluate both the inter-cluster and intra-cluster distance in hierarchical clustering(E-cluster), and use the sum of squares of deviations(SSD) as a regularization term to further balance the diversity and similarity of energy distance evaluation. We evaluate our method on large scale re-ID datasets, including Market-1501, DukeMTMC-reID and MARS. Extensive experiments show that our method obtains significant improvements over the state-of-the-art unsupervised methods, and even better than some transfer learning methods.• We measure the distance between clusters with energy distance, which can promote more compact clustering.

Threshold-Based Hierarchical Clustering for Person Re-Identification

“…Person re-identification(re-ID) is a practical task about finding a specific person from cameras and it is widely used in security fields. Unsupervised domain adaptation (UDA) re-ID [ 1 , 2 , 3 ] has been popular because it does not require a lot of labeled data with respect to supervised methods. UDA re-ID requires labeled source datasets and unlabeled target datasets.…”

Exploiting robust unsupervised video person re‐identification

“…In the context of unsupervised learning, approaches for reID can be roughly divided into four categories: (1) transfer learning-based methods [7][8][9]; (2) one-shot learning-based methods [10][11][12]; (3) clustering-based methods [13][14][15]; (4) tracklet-based association learning [16][17][18][19]. These methods above employed global-level features to achieve unsupervised learning.…”

A Joint 2D-3D Complementary Network for Stereo Matching

“…Stereo matching is the progress of getting the depth information from stereo image pairs in the same scene, which is essential for Autonomous Driving [ 1 ], 3D Reconstruction and Mapping [ 2 ], Human-Computer Interaction [ 3 ], Marine Science and Systems [ 4 ], Planetary Exploration [ 5 ], Unmanned Aerial Vehicles (UAV) [ 6 ] or Person Re-identification [ 7 , 8 ]. Compared with expensive lidar equipment, stereo matching is convenient and high-efficient.…”