Abstract:Person re-identification (re-id) is a cross-camera retrieval task which establishes a correspondence between images of a person from multiple cameras. Deep Learning methods have been successfully applied to the problem and achieved impressive results. However, these methods require large amounts of labeled training data. Current labeled datasets in person re-id are limited in scale and manually acquiring such large-scale dataset in surveillance camera is a tedious and labor-intensive task. In this paper, we propose a semi-supervised framework that performs Sparse Label Smoothing Regularization (SLSR) by considering similarities between unlabeled sample and training sample in the same feature space. Our approach first exploits the clustering property of existing person re-id datasets to create groups of similar objects that model the correlation among view. We make use of the training set to create cluster of similar objects using the intermediate feature representation of a CNN model. Each cluster is used to generate synthetic data samples using a generative adversarial model. We finally defined a sparse smoothing regularization term and train the network with join supervision of cross-entropy loss. The proposed approach tackles two problems (1) how to efficiently use the generated data and (2) how to address the over-smoothness problem found in current regularization. We solve these two problems by using a generative model for data augmentation and by maintaining and propagating similarities across the network through the concatenation of training images and generated images into one homogeneous feature space. Extensive experiments on four large-scale datasets show that our regularization method significantly improves the Re-ID accuracy compared to existing semi-supervised methods.

Generated Data With Sparse Regularized Multi-Pseudo Label for Person Re-Identification

Online Pseudo Label Generation by Hierarchical Cluster Dynamics for Adaptive Person Re-identification

Sparse Label Smoothing Regularization for Person Re-Identification

Reasonably Assign Label Distributions to GAN Images in Person Re-identification baseline

Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in Vitro

Feature Affinity-Based Pseudo Labeling for Semi-Supervised Person Re-Identification

Sparse Label Smoothing for Semi-supervised Person Re-Identification.

Cross-dataset Person Re-Identification Using Similarity Preserved Generative Adversarial Networks.

Generating Pedestrian Images For Person Re-Identification

Joint Discriminative and Generative Learning for Person Re-Identification

A GAN-Based Self-Training Framework for Unsupervised Domain Adaptive Person Re-Identification

Real-World Person Re-Identification Via Super-Resolution and Semi-Supervised Methods

Learning Generic Feature Representations with Adversarial Regularization for Person Re-Identification.

Style Uncertainty Based Self-Paced Meta Learning for Generalizable Person Re-Identification.

An Improved CycleGAN for Data Augmentation in Person Re-Identification.

Data Generation for Improving Person Re-identification

Pseudo-Label Noise Prevention, Suppression and Softening for Unsupervised Person Re-Identification.

A Unified Generative Adversarial Framework for Image Generation and Person Re-identification.

Center Based Pseudo-Labeling For Semi-Supervised Person Re-Identification

Semantic Map Guided Identity Transfer GAN for Person Re-identificationJust Accepted

Semi-supervised Learning for Person Re-Identification Based on Style-Transfer-generated Data by CycleGANs