Attention-Aware Compositional Network for Person Re-identification

Abstract

Person re-identification (ReID) is to identify pedestrians observed from different camera views based on visual appearance. It is a challenging task due to large pose variations, complex background clutters and severe occlusions. Recently, human pose estimation by predicting joint locations was largely improved in accuracy. It is reasonable to use pose estimation results for handling pose variations and background clutters, and such attempts have obtained great improvement in ReID performance. However, we argue that the pose information was not well utilized and hasn't yet been fully exploited for person ReID.In this work, we introduce a novel framework called Attention-Aware Compositional Network (AACN) for person ReID. AACN consists of two main components: Poseguided Part Attention (PPA) and Attention-aware Feature Composition (AFC). PPA is learned and applied to mask out undesirable background features in pedestrian feature maps. Furthermore, pose-guided visibility scores are estimated for body parts to deal with part occlusion in the proposed AFC module. Extensive experiments with ablation analysis show the effectiveness of our method, and stateof-the-art results are achieved on several public datasets, including Market-1501, CUHK03, CUHK01, SenseReID, CUHK03-NP and DukeMTMC-reID.

ABD-Net: Attentive but Diverse Person Re-Identification

“…Motivated by such observations, the attention mechanism [7] was introduced to enforce the features to mainly capture the discriminative appearances of human bodies (or certain body parts). Since then, the attention-based models [8,9,10,11,12] have boosted person Re-ID performance much.…”

Beyond Human Parts: Dual Part-Aligned Representations for Person Re-Identification

“…Similar to the semantic segmentation approaches, poses or keypoints estimation can also be used for accurate/reliable human part localization. For example, there are approaches exploring both the human poses and the human part masks [9], or generating human part masks via exploting the connectivity of the keypoints [50]. There are some other studies [5,29,35,55] that also exploit the pose cues to extract the part-aligned features.…”

“…Attention for ReID. Attention mechanisms have been used to capture human part information in recent work [21,56,50,17,34]. Typically, the predicted attention maps distribute most of the attention weights on human parts that may help improve the results.…”

Weakly-Supervised Completion Moment Detection using Temporal Attention

“…weak labels), through utilising temporal attention learning. Attention Learning has proven beneficial for research problems, such as image captioning [40,41,6], object detection and tracking [3,7,1] and person reidentification [12,20,39]. Recently action recognition and localisation have also used attention networks to learn which spatial and/or temporal regions contain the most discriminative information.…”