Embedding Deep Metric for Person Re-identification: A Study Against Large Variations

By Hailin Shi, Yang Yang, Xiangyu Zhu, Shengcai Liao, Zhen Lei, Weishi Zheng, S. Li. European Conference on Computer Vision, 2016

313 citations recorded by the source corpus.

Person re-identification is challenging due to the large variations of pose, illumination, occlusion and camera view. Owing to these variations, the pedestrian data is distributed as highly-curved manifolds in the feature space, despite the current convolutional neural networks (CNN)’s capability of feature extraction. However, the distribution is unknown, so it is difficult to use the geodesic distance when comparing two samples. In practice, the current deep embedding methods use the Euclidean distance for the training and test. On the other hand, the manifold learning methods suggest to use the Euclidean distance in the local range, combining with the graphical relationship between samples, for approximating the geodesic distance. From this point of view, selecting suitable positive (i.e. intra-class) training samples within a local range is critical for training the CNN embedding, especially when the data has large intra-class variations. In this paper, we propose a novel moderate positive sample mining method to train robust CNN for person re-identification, dealing with the problem of large variation. In addition, we improve the learning by a metric weight constraint, so that the learned metric has a better generalization ability. Experiments show that these two strategies are effective in learning robust deep metrics for person re-identification, and accordingly our deep model significantly outperforms the state-of-the-art methods on several benchmarks of person re-identification. Therefore, the study presented in this paper may be useful in inspiring new designs of deep models for person re-identification.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/fc448f339e0576311954caad66d731496ef516a1

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

In Computer Vision and Pattern Recognition

Citations in the corpus

Deep Representation Learning With Part Loss for Person Re-Identification (2017, background): “There exist several works [24], [25], [35]–[38] that use siamese network to test whether the two input images contain the same person.”
Deep Spatial-Semantic Attention for Fine-Grained Sketch-Based Image Retrieval (2017, methodology): “Mahalanobis distance [41, 35] is another example of a higher-order energy function in that it does O(N(2)) comparisons for N channels.”
In Defense of the Triplet Loss for Person Re-Identification (2017, background): “However, using pretrained networks also leads to a design lock-in, and does not allow for the exploration of new deep learning advances or different architectures.”
Person Re-identification by Deep Learning Multi-scale Representations (2017, background): “Recently, deep learning re-id models [26, 4, 48, 56, 44, 77, 64, 7, 11, 65, 9, 60, 27] start to take over and have obtained impressive performance.”
Pose-Driven Deep Convolutional Model for Person Re-identification (2017, methodology): “Predefined rigid body parts are also used by many deep learning based methods [5, 47, 40] for the purpose of learning local pedestrian features.”
Hard-Aware Deeply Cascaded Embedding (2016, background): “Deep metric embedding has attracted increasing attention for various tasks , such as visual product search [6, 18, 23, 27, 35], face recognition [21, 32, 41, 4], local image descriptor learning [9, 2, 12, 24], person/vehicle reidentification [22, 7, 43, 17], zero-shot image classification [19, 45, 5”
Multi-scale Deep Learning Architectures for Person Re-identification (2017, methodology): “Recently, inspired by the success of convolutional neural networks (CNN) in many computer vision problems, deep CNN architectures [1, 48, 27, 51, 44, 50, 4] have been widely used for person re-id.”
SVDNet for Pedestrian Retrieval (2017, background): “The former is also called deep metric learning, in which image pairs or triplets are used as input to the network [25, 24, 1, 13, 5, 19].”
Learning Deep Context-Aware Features over Body and Latent Parts for Person Re-identification (2017, methodology): “CUHK03: For the CUHK03 dataset, we compare our method with many existing approaches, including Filter Pair Neural Networks (FPNN) [20], Improved Deep Learning Architecture (IDLA) [1], Cross-view Quadratic Discriminant Analysis (XQDA) [22], PSD constrained asymmetric metric learning (denoted as MLAPG”
Deep Learning for Person Re-Identification: A Survey and Outlook (2020, background): “The basic idea is to select the triplets which are hard for the model to discriminate [48], [153].”
Scalable Person Re-identification on Supervised Smoothed Manifold (2017, background): “Motivated by the fact that pedestrian data are distributed on a highly curved manifold, a sampling strategy for training neural network called Moderate Positive Mining (MPM) is proposed in [44].”
Deep Transfer Learning for Person Re-Identification (2016, methodology): “Deep Re-ID models Existing deep Re-ID models [7], [8], [13]–[15] differ significantly in their network architectures, which are largely determined by the training objectives/losses.”
Deep Metric Learning with Hierarchical Triplet Loss (2018, background): “…single-shot object classiﬁcation [30,34,32], key-point descriptor learning [12,24], face veriﬁcation [22,20], person re-identiﬁcation [30,23], object tracking [29] and etc. Recently, there is a number of widely-used loss functions developed for deep metric learning, such as contrastive…”