Attention-Aware Compositional Network for Person Re-identification

By Jing Xu, Rui Zhao, Feng Zhu, Huaming Wang, Wanli Ouyang. 2018

468 citations recorded by the source corpus.

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: Pose-guided 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 state-of-the-art results are achieved on several public datasets, including Market-1501, CUHK03, CUHK01, SenseReID, CUHK03-NP and DukeMTMC-reID.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Objects as Points (2019, background): “It has down-stream applications in surveillance [57], autonomous driving [53], and visual question answering [1].”
Knowledge Distillation for End-to-End Person Search (2019, unknown): “More recent re-identiﬁcation approaches are based on CNNs [1, 31, 62, 64] and mainly concerned with the estimation of a ID-feature embedding spaces, either via Siamese networks and contrastive losses [1, 9, 11, 15, 31, 36, 50, 56, 62], or with ID-classiﬁcation and cross-entropy losses [54, 67].”
Omni-Scale Feature Learning for Person Re-Identification (2019, methodology): “In [66, 40, 57, 42, 47, 65], body part-specific CNNs are learned by means of off-the-shelf pose detectors.”
Relation-Aware Global Attention for Person Re-Identification (2019, background): “Recently, many studies resort to attention design to address the aforementioned challenges in person re-id by strengthening the discriminative features and suppressing interference [22, 39, 20, 12, 11, 6].”
Hierarchical Clustering With Hard-Batch Triplet Loss for Person Re-Identification (2019, background): “With the development of convolutional neural networks (CNN), supervised re-ID [9, 11, 21, 25, 30, 2, 31, 23] has achieved excellent performance.”
Each Part Matters: Local Patterns Facilitate Cross-View Geo-Localization (2020, methodology): “learning on the convolutional layer, but is different in two aspects: Different from works of the first line [37], [36], [48], [49], [50], the proposed method does not need an extra part estimator.”
Diverse Part Discovery: Occluded Person Re-identification with Part-Aware Transformer (2021, methodology): “Existing ReID methods can be summarized to hand-crafted descriptors [47, 23], metric learning methods [56, 20, 24] and deep learning methods[38, 25, 33, 35, 45, 52, 19, 21, 34, 26, 22].”
Improving Person Re-identification by Attribute and Identity Learning (2017, background): “25 AACN[32] CVPR2018 GoogLeNet 85.”
Interaction-And-Aggregation Network for Person Re-Identification (2019, methodology): “It is noted that: (1) The gaps between our results and the spatial alignment methods (Spindle [59], AACN [54] and PSE [35]) that incorporate an external part detection sub-network are signiﬁcant: about 10% improvement on top-1 accuracy and mAP.”
Attention mechanisms in computer vision: A survey (2021, unknown): “[8], [9], [15], [21], [31], [32], [34], [35] , [22], [26], [62], [63], [64], [65], [66], [67] , [41], [68], [69], [70], [71], [72], [73], [74] , [8], [34], [42], [43], [75], [76], [77], [78] , [27], [44], [45], [46], [79], [80], [81], [82] , [61], [83], [84], [85], [86], [87], [88], [89] , [47], [90”
A Strong Baseline and Batch Normalization Neck for Deep Person Re-Identification (2019, methodology): “Attention-based methods [25]–[28] involve an attention mechanism to extract additional discriminative features.”
Region Graph Embedding Network for Zero-Shot Learning (2020, methodology): “More recently, by pursing automatic part discovery [53], attention mechanisms [57, 56, 55, 25] have been applied into ZSL and GZSL [52, 80, 78, 30] for capturing multiple semantic regions, which can facilitate desirable knowledge transfer.”
Robust Person Re-Identification by Modelling Feature Uncertainty (2019, unknown): “To overcome misalignment and pose variations, several works use pose detectors to identify body parts on which part-specific features are learned [34, 25, 31, 32].”
ABD-Net: Attentive but Diverse Person Re-Identification (2019, methodology): “Second, our attention masks are directly learned from the data and context, without relying on manually-defined parts, part region proposals, nor pose estimation [8, 9, 11].”
Identity-Guided Human Semantic Parsing for Person Re-Identification (2020, background): “Many works employ extra semantic in terms of part/pose to locate body parts [6, 27, 29, 30, 32, 34, 40, 46, 49, 56] and try to achieve the pixel-level alignment.”