Person Re-identification by Attributes

By Ryan Layne, Timothy M. Hospedales, Shaogang Gong. 2012

344 citations recorded by the source corpus.

Visually identifying a target individual reliably in a crowded environment observed by a distributed camera network is critical to a variety of tasks in managing business information, border control, and crime prevention. Automatic re-identification of a human candidate from public space CCTV video is challenging due to spatiotemporal visual feature variations and strong visual similarity between different people, compounded by low-resolution and poor quality video data. In this work, we propose a novel method for re-identification that learns a selection and weighting of mid-level semantic attributes to describe people. Specifically, the model learns an attribute-centric, parts-based feature representation. This differs from and complements existing low-level features for re-identification that rely purely on bottom-up statistics for feature selection, which are limited in discriminating and identifying reliably visual appearances of target people appearing in different camera views under certain degrees of occlusion due to crowdedness. Our experiments demonstrate the effectiveness of our approach compared to existing feature representations when applied to benchmarking datasets.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

FairFace: Face Attribute Dataset for Balanced Race, Gender, and Age (2019, background): “Attributes are also widely used for person re-identification in images or videos, combining features from human face and body appearance [33, 34, 53], especially effective when faces are not fully visible or too small.”
Faceless Person Recognition: Privacy Implications in Social Media (2016, background): “Various other cues have been explored, for example: attributes classification [40, 41], social context [42, 43], relative camera positions [44], space-time priors [45], and photo-album priors [46].”
Person Recognition in Personal Photo Collections (2015, background): “Multiple other cues have been explored, for example: attributes classification [24, 25], explicit cloth modelling [15], relative camera positions [18], social context [16, 35], space-time priors [28], and photo-album priors [34].”
Bilinear Attention Networks for Person Retrieval (2019, background): “Other solutions benefit from person attributes [16, 35], which are invariant across pose, illumination, misalignment etc.”
Deep Attributes Driven Multi-Camera Person Re-identification (2016, methodology): “7 Traditional Attributes Learning based ReID AIR [27] 18.”
FairFace: Face Attribute Dataset for Balanced Race, Gender, and Age for Bias Measurement and Mitigation (2021, methodology): “Face attribute recognition has been applied as a subcomponent to other computer vision tasks such as face verification [31] and person re-idenfication [33, 34, 55].”
Clothing Attributes Assisted Person Reidentification (2015, methodology): “[20], [21] on person reidentification is the most related one to our approach.”
Transferable Joint Attribute-Identity Deep Learning for Unsupervised Person Re-identification (2018, methodology): “Attribute for Re-ID Visual attributes have been exploited as a mid-level feature representation for cross-view re-id [24, 22, 23, 43, 42, 36].”
Deep domain adaptation for describing people based on fine-grained clothing attributes (2015, background): “Predicting human attributes [28, 1, 47, 11] is important for many surveillance applications, such as person re-identification across cameras [30], suspect search based on eyewitness”
Attribute-Aware Attention Model for Fine-grained Representation Learning (2018, methodology): “rt-based model for each category and a cascading scheme for fine-grained classification. Attributes for Image Recognition. In image recognition, attribute is a type of important auxiliary information [17, 18] and have been investigated in many works. Yutian et al. [22], Tetsu et al. [26] train CNNs ”
Pedestrian Attribute Recognition At Far Distance (2014, methodology): “However, when large variations are presented, one may consider modeling the local smoothness and discovering the intrinsic manifold of the data.”
Hierarchical Gaussian Descriptor for Person Re-identification (2016, background): “Attribute based descriptors obtain lingual description of person images [17].”
Multi-Task Learning with Low Rank Attribute Embedding for Person Re-Identification (2015, background): “For person re-identification, attributes are powerful in preserving consistent representations of the same person and capturing differences among different people [29, 26, 27, 28].”
Cross-Domain Image Retrieval with a Dual Attribute-Aware Ranking Network (2015, background): “In the surveillance domain, matching clothing images across cameras is a fundamental task for the well-known person re-identification problem [28, 37].”
Cumulative Attribute Space for Age and Crowd Density Estimation (2013, methodology): “Attributes have been successfully applied for solving various computer vision problems by classification [11, 20, 22, 24], but have never been used for regression to the best of our knowledge.”
Improving Person Re-identification by Attribute and Identity Learning (2017, methodology): “In [17, 16, 15], low-level descriptors and SVM is used to train attribute detectors, and the attributes are integrated in several metric learning methods.”