Feature Erasing and Diffusion Network for Occluded Person Re-Identification

  title={Feature Erasing and Diffusion Network for Occluded Person Re-Identification},
  author={Zhikang Wang and Feng Zhu and Shixiang Tang and Rui Zhao and Lihuo He and Jiangning Song},
  journal={2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
Occluded person re-identification (ReID) aims at matching occluded person images to holistic ones across different camera views. Target Pedestrians (TP) are often disturbed by Non-Pedestrian Occlusions (NPO) and Non-Target Pedestrians (NTP). Previous methods mainly focus on increasing the model's robustness against NPO while ignoring feature contamination from NTP. In this paper, we propose a novel Feature Erasing and Diffusion Network (FED) to simultaneously handle challenges from NPO and NTP… 

Figures and Tables from this paper

Dynamic Feature Pruning and Consolidation for Occluded Person Re-Identification

A feature pruning and consolidation framework to circumvent explicit human structure parse and outperforms state-of-the-art results by at least 8.6% mAP and 6.0% Rank-1 accuracy on the challenging Occluded-Duke dataset.

Parallel Augmentation and Dual Enhancement for Occluded Person Re-identification

This work designs a parallel augmentation mechanism for Occluded Re-ID to generate more suitable occluded data to mitigate the neg-ative effects of unbalanced data and proposes the dual enhancement strategy for global and local features to promote the context information and details.

Deep Learning-based Occluded Person Re-identification: A Survey

A systematic survey of occluded person Re-ID methods from the perspective of issues and solutions and insights on promising future research directions are provided.

Body Part-Based Representation Learning for Occluded Person Re-Identification

GiLt is proposed, a novel training scheme for learning part-based representations that is robust to occlusions and non-discriminative local appearance and outperforms state-of-the-art methods on the challenging Occluded-Duke dataset.

Relation Preserving Triplet Mining for Stabilising the Triplet Loss in Re-identification Systems

Relation Preserving Triplet Mining (RPTM) is introduced, a feature matching guided triplet mining scheme that ensures that triplets will respect the natural subgroupings within an object ID.

Multiplex-detection Based Multiple Instance Learning Network for Whole Slide Image Classification

This paper proposes a novel multiplex-detection-based multiple instance learning (MDMIL) and adopts a memory-based contrastive loss, which is practicable for WSI classification with a single input of each iteration.



Occlude Them All: Occlusion-Aware Attention Network for Occluded Person Re-ID

The Occlusion-Aware Mask Network (OAMN) is proposed, which incorporates an attention-guided mask module, which requires guidance from labeled occlusion data, and offers a novel occluded augmentation scheme that suits real-world scenarios better than existing schemes, which consider only limited types of occlusions.

Neighbourhood-guided Feature Reconstruction for Occluded Person Re-Identification

This work proposes to reconstruct the feature representation of occluded parts by fully exploiting the information of its neighborhood in a gallery image set by introducing a visible part-based feature by body mask for each person image.

Foreground-Aware Pyramid Reconstruction for Alignment-Free Occluded Person Re-Identification

An alignment-free matching approach namely Foreground-aware Pyramid Reconstruction (FPR) is developed to accurately compute matching scores between occluded persons, regardless of their different scales and sizes and a occlusion-sensitive foreground probability generator is designed that focuses more on clean human body parts to robustify the similarity computation with less contamination from occlusions.

Attention-Aware Compositional Network for Person Re-identification

This work introduces a novel framework called Attention-Aware Compositional Network (AACN) for person ReID, which consists of two main components: Pose-guided Part Attention (PPA) and Attention-aware Feature Composition (AFC).

Human Parsing Based Alignment With Multi-Task Learning For Occluded Person Re-Identification

A model named HPNet is proposed to extract part-level features and predict visibility of each part, based on human parsing, which surpasses previous methods by a large margin under occlusion scenarios.

Occluded Person Re-Identification

This paper proposes a novel Attention Framework of Person Body (AFPB) based on deep learning, consisting of an Occlusion Simulator (OS) which automatically generates artificial occlusions for full-body person images, and multi-task losses that force the neural network not only to discriminate a person's identity but also to determine whether a sample is from the occluded data distribution or the full- body data distribution.

Attention Driven Person Re-identification

Guided Saliency Feature Learning for Person Re-identification in Crowded Scenes

A new matching approach, called Guided Adaptive Spatial Matching (GASM), which expects that each spatial feature in the query can find the most similar spatial features of a person in a gallery to match.

Pose-Guided Visible Part Matching for Occluded Person ReID

A Pose-guided Visible Part Matching (PVPM) method that jointly learns the discriminative features with pose-guided attention and self-mines the part visibility in an end-to-end framework that achieves competitive performance to state-of-the-art methods.

High-Order Information Matters: Learning Relation and Topology for Occluded Person Re-Identification

  • Guan'an WangShuo Yang Jian Sun
  • Computer Science
    2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
  • 2020
This paper proposes a novel framework by learning high-order relation and topology information for discriminative features and robust alignment for occluded person re-identification and significantly outperforms state-of-the-art mAP scores on Occluded-Duke dataset.