Deep Metric Learning for Person Re-identification

By Dong Yi, Zhen Lei, Shengcai Liao, S. Li. 2014

900 citations recorded by the source corpus.

Subfields: Biomedical Engineering, 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 Biomedical Engineering

1. Yi Yang
2. Liang Zheng
3. Qi Tian
4. Zhedong Zheng
5. Shiliang Zhang
6. Tao A. Xiang
7. Wen Gao
8. G. Wang

In Computer Vision and Pattern Recognition

1. Yi Yang
2. Luc Van Gool
3. Dacheng Tao
4. Xiaogang Wang
5. Liang Zheng
6. Ming-Hsuan Yang
7. Qi Tian
8. Bernt Schiele

Citations in the corpus

• Person Re-identification by Multi-Channel Parts-Based CNN with Improved Triplet Loss Function (2016, methodology): “Yi et al. [45] constructed a siamese neural network (denoted as DeepM in our paper) to learn pairwise similarity, and also used body parts to train their CNN models.”
• Pose-Guided Feature Alignment for Occluded Person Re-Identification (2019, background): “Recent years, deep learning methods [39, 34, 13, 27, 3, 25, 7, 6, 36, 21] for person re-id show significant superiority on retrieval accuracy.”
• Camera Style Adaptation for Person Re-identification (2017, methodology): “Many deep learning methods [38, 34, 33, 3, 24] have been proposed in person re-ID.”
• Deep Learning for Person Re-Identification: A Survey and Outlook (2020, background): “Extensive models have been developed to handle the various challenges, concentrating on feature representation learning [45], [46], distance metric learning [47], [48] or their combinations.”
• MARS: A Video Benchmark for Large-Scale Person Re-Identification (2016, methodology): “Image pairs [1,23,40] or triplets [8] are defined as input to CNN, instead of single training images.”
• GLAD: Global-Local-Alignment Descriptor for Pedestrian Retrieval (2017, methodology): “Person Re-ID is commonly solved as a classification or distance metric learning problem [1, 15, 18, 24, 27, 36, 40, 41], which suffer from massive computations of classifier training and the high time complexity for online pairwise matching.”
• Knowledge Distillation for End-to-End Person Search (2019, neutral): “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].”
• Learning Discriminative Features with Multiple Granularities for Person Re-Identification (2018, background): “[20, 42] first introduce deep siamese network architecture into Re-ID and combine the body part feature learning, achieving higher performances comparing to the contemporary hand-crafted meth-ods.”
• DSSL: Deep Surroundings-person Separation Learning for Text-based Person Retrieval (2021, background): “Current researches of person retrieval chiefly focus on image-based person retrieval [7, 27, 29] (aka. person re-identification), which may sometimes suffer from lacking query images of the target pedestrian in practical application.”
• Visual Attention Methods in Deep Learning: An In-Depth Survey (2022, background): “to tackle the limitation of the previous attention modules in person Re-Identification by learning attention selection and feature representation jointly and hence solving the issue of misalignment calibration caused by constrained attention mechanisms [33], [34], [35], [36].”
• Deep Representation Learning With Part Loss for Person Re-Identification (2017, background): “[37] jointly learn the color feature, texture descriptor, and distance metric in a siamese deep neural network.”
• Beyond Triplet Loss: A Deep Quadruplet Network for Person Re-identification (2017, methodology): “In addition to a triplet loss, some deep learning methods [19, 1, 41, 35, 29, 32] address the person ReID problem from the classification aspect and adopt a binary classification loss to train their models.”
• Deep Metric Learning: A Survey (2019, background): “It is also possible to combine the Siamese network and Convolutional neural network, which has important advantages [97], which include similarity learning from direct image pixels, color and texture information at the same time, and its flexible structure.”
• Pedestrian Alignment Network for Large-scale Person Re-Identification (2017, methodology): “[13] split a pedestrian image into three horizontal parts and respectively trained three part-CNNs to extract features.”
• Jointly Attentive Spatial-Temporal Pooling Networks for Video-Based Person Re-identification (2017, background): “tation learning [11, 19, 15, 38], distance metric learning [14, 41, 30, 36, 37, 21] and CNN-based schemes [33, 32, 24, 25].”
• Attribute-Aware Attention Model for Fine-grained Representation Learning (2018, background): “g methods extract hand-crafted features [20, 51], especially color and texture descriptors, then use distance metric learning [14, 20] to further capture discriminative factors. Deep learning methods [1, 36, 46] learn image representations and distance metrics simultaneously in an end-to-end manner.”
• Horizontal Pyramid Matching for Person Re-identification (2018, background): “[35] first employed deep neural network to determine if a pair of input images belong to the same ID.”
• Mixed High-Order Attention Network for Person Re-Identification (2019, background): “Person ReID & Attention Mechanism: Person ReID intends to correctly match images of pedestrians across videos captured from different cameras, it has been widely studied, such as ranking by pairwise constraints [34, 48], metric learning [54, 51], deep embedding learning [63, 43], re-ranking [62, 16]”