Deep Metric Learning for Practical Person Re-Identification

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

170 citations recorded by the source corpus.

Various hand-crafted features and metric learning methods prevail in the field of person re-identification. Compared to these methods, this paper proposes a more general way that can learn a similarity metric from image pixels directly. By using a "siamese" deep neural network, the proposed method can jointly learn the color feature, texture feature and metric in a unified framework. The network has a symmetry structure with two sub-networks which are connected by Cosine function. To deal with the big variations of person images, binomial deviance is used to evaluate the cost between similarities and labels, which is proved to be robust to outliers. Compared to existing researches, a more practical setting is studied in the experiments that is training and test on different datasets (cross dataset person re-identification). Both in "intra dataset" and "cross dataset" settings, the superiorities of the proposed method are illustrated on VIPeR and PRID.

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

In Computer Vision and Pattern Recognition

Citations in the corpus

DISCO: Dynamic and Invariant Sensitive Channel Obfuscation for deep neural networks (2020, methodology): “While sophisticated learning based partitioning methods have been explored [36, 37, 38, 39], methods such as [40] have achieved outstanding performance with trivial deterministic splitting.”
Faceless Person Recognition: Privacy Implications in Social Media (2016, background): “Like other areas of computer vision, the last years have seen a shift from classifiers based on handcrafted features and metric learning approaches [13, 14, 15, 16, 17, 18, 19] towards methods based on deep learning [20, 21, 22, 23, 24, 25, 26, 27].”
Learning Deep Disentangled Embeddings with the F-Statistic Loss (2018, methodology): “…between researchers focusing on deep embeddings which are evaluated via few-shot learning (e.g., Ustinova & Lempitsky, 2016; Schroff et al., 2015; Yi et al., 2014b) and researchers focusing on few-shot learning who have found deep embeddings to be a useful method (e.g., Snell et al., 2017;…”
Beyond frontal faces: Improving Person Recognition using multiple cues (2015, methodology): “[37] used a siamese deep network to learn the similarity metric between pairs of images.”
Person Recognition in Personal Photo Collections (2015, background): “Both settings have seen a recent shift from sophisticated classifiers based on hand-crafted features and metric learning approaches [20, 7, 5, 29, 26, 41, 1], towards methods based on deep learning [37, 36, 43, 33, 27, 38, 21].”
Embedding Deep Metric for Person Re-identification: A Study Against Large Variations (2016, methodology): “The re-identification methods by deep learning [1,6,17,35] incorporate the two above-mentioned aspects (feature extraction and metric learning) of person re-identification into an integrated framework.”
Recurrent Convolutional Network for Video-Based Person Re-identification (2016, methodology): “In the area of image based person re-identification, DNNs have been used to learn ranking functions based on pairs [45], or triplets of images [5].”
Recurrent Attention Models for Depth-Based Person Identification (2016, methodology): “Second, the methods above [73, 41, 19, 1] ingest several images as input and compute similarity between these inputs.”
Deep Representation Learning With Part Loss for Person Re-Identification (2017, background): “There exist several works [25], [34], [24], [35], [36], [37] that use siamese network to test whether the two input images contain the same person.”
Deep Attributes Driven Multi-Camera Person Re-identification (2016, methodology): “Some previous works like DML [41], IDLA [42] and Deep-RDC [24] take advantages of deep learning in person ReID.”
Multi-Similarity Loss With General Pair Weighting for Deep Metric Learning (2019, methodology): “Siamese network [6] is a representative pair-based method that learns an embedding via contrastive loss.”
Deep Relative Distance Learning: Tell the Difference between Similar Vehicles (2016, background): “Yi et al.[22] applied a “siamese” deep network which has a symmetric structure with two sub-network to learn pair-wise similarity.”
Deep Metric Learning with Angular Loss (2017, background): “It was originally designed for signature verification [4], but gained a lot of attention recently due to its superior performance in face verification [6, 30, 28, 36].”
Hyperbolic Image Embeddings (2019, methodology): “Alternatively, some few-shot learning [53], face recognition [41], and person re-identification methods [52, 59]”
An improved deep learning architecture for person re-identification (2015, methodology): “The deepmetric-learning-based method [27] also reports results on the VIPeR data set, with a lower rank-1 accuracy of 28.”
Pose-Driven Deep Convolutional Model for Person Re-identification (2017, methodology): “When handling a pair of person images, existing deep learning methods usually learn feature representations of each person by using a deep matching function from convolutional features [1, 25, 53, 13] or from the Fully Connected (FC) features [58, 40, 61].”
Learning Deep Embeddings with Histogram Loss (2016, methodology): “In particular, we have evaluated the Binomial Deviance loss [27].”
A Discriminatively Learned CNN Embedding for Person Reidentification (2016, background): “Given a label s ∈ { 0 , 1 } , the veriﬁ-cation network forces two images of the same person to be mapped to nearby points in the feature space.”
Person Recognition in Personal Photo Collections (2017, background): “There is an active line of research on pedestrian re-id, starting with hand-crafted features [30], [31], [32] and has moved towards deep feature based schemes [28], [33], [34], [35], [36], [37], [38], [39], [40].”
Adapted Deep Embeddings: A Synthesis of Methods for k-Shot Inductive Transfer Learning (2018, methodology): “1 Various objective functions have been proposed for deep metric learning, all of which aim to ensure that instances of the same class are near one another in the embedding space and instances of different classes are far apart [1, 3, 17, 19, 24, 26, 28, 31, 33, 34, 37].”