Viewpoint-Aware Loss with Angular Regularization for Person Re-Identification

Abstract

Although great progress in supervised person re-identification (Re-ID) has been made recently, due to the viewpoint variation of a person, Re-ID remains a massive visual challenge. Most existing viewpoint-based person Re-ID methods project images from each viewpoint into separated and unrelated sub-feature spaces. They only model the identity-level distribution inside an individual viewpoint but ignore the underlying relationship between different viewpoints. To address this problem, we propose a novel approach, called Viewpoint-Aware Loss with Angular Regularization (VA-reID). Instead of one subspace for each viewpoint, our method projects the feature from different viewpoints into a unified hypersphere and effectively models the feature distribution on both the identity-level and the viewpoint-level. In addition, rather than modeling different viewpoints as hard labels used for conventional viewpoint classification, we introduce viewpoint-aware adaptive label smoothing regularization (VALSR) that assigns the adaptive soft label to feature representation. VALSR can effectively solve the ambiguity of the viewpoint cluster label assignment. Extensive experiments on the Market1501 and DukeMTMC-reID datasets demonstrated that our method outperforms the state-of-the-art supervised Re-ID methods.

Unsupervised Person Re-Identification via Multi-Order Cross-View Graph Adversarial Network

“…At present, most proposed person re-id models are under supervised framework to explore distance metric [32]- [34], [50], view-invariant discriminative feature [2], [27], [51] or deep learning [11], [14]. Zhu et al [50] proposed a hard and easy negative samples mining based distance learning approach for person re-identification, which learns the distance metric by designing different objective functions for hard and easy negative samples.…”

Multi-Camera Trajectory Forecasting with Trajectory Tensors

“…For example, on the commonly used Market [25] benchmark, rank 1 person RE-ID performance (i.e. correct RE-ID given 1 guess only) has improved from 47.3% in 2015 [25] to 96.8% in 2020 [26]. Most work on person RE-ID has focused on image-level matching, and current state-of-the-art methods [27], [28] exploit this visual cue without other sources of information.…”

Angular regularization for unsupervised domain adaption on person re-identification

“…In face recognition researches, some variants of SoftMax loss have be proposed to learn the metric space based on angular margin to enhance the discriminative power [14][15][16][17]. In person Re-ID field, some researchers studied the angular regularization on supervised [18][19][20] or cross-domain [21] Re-ID task. Also the angular regularization is introduced into clustering-based UDA Re-ID model to compensate for the weakness of SoftMax loss [7].…”