SphereReID: Deep Hypersphere Manifold Embedding for Person Re-Identification

Abstract

Many current successful Person Re-Identification(ReID) methods train a model with the softmax loss function to classify images of different persons and obtain the feature vectors at the same time. However, the underlying feature embedding space is ignored. In this paper, we use a modified softmax function, termed Sphere Softmax, to solve the classification problem and learn a hypersphere manifold embedding simultaneously. A balanced sampling strategy is also introduced. Finally, we propose a convolutional neural network called SphereReID adopting Sphere Softmax and training a single model end-to-end with a new warming-up learning rate schedule on four challenging datasets including Market-1501, DukeMTMC-reID, CHHK-03, and CUHK-SYSU. Experimental results demonstrate that this single model outperforms the state-of-the-art methods on all four datasets without fine-tuning or re-ranking. For example, it achieves 94.4% rank-1 accuracy on Market-1501 and 83.9% rank-1 accuracy on DukeMTMC-reID. The code and trained weights of our model will be released.

arXiv commentContribute to Journal of Visual Communication and Image Representation

References in corpus (17)

• In Defense of the Triplet Loss for Person Re-Identification
• Additive Margin Softmax for Face Verification
• End-to-End Comparative Attention Networks for Person Re-identification
• Deep Representation Learning with Part Loss for Person Re-Identification
• AlignedReID: Surpassing Human-Level Performance in Person Re-Identification
• GLAD: Global-Local-Alignment Descriptor for Pedestrian Retrieval
• SphereFace: Deep Hypersphere Embedding for Face Recognition
• Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in vitro
• Performance Measures and a Data Set for Multi-Target, Multi-Camera Tracking
• Harmonious Attention Network for Person Re-Identification
• Deeply-Learned Part-Aligned Representations for Person Re-Identification
• SVDNet for Pedestrian Retrieval
• Beyond triplet loss: a deep quadruplet network for person re-identification
• Gated Siamese Convolutional Neural Network Architecture for Human Re-Identification
• A Siamese Long Short-Term Memory Architecture for Human Re-Identification
• DarkRank: Accelerating Deep Metric Learning via Cross Sample Similarities Transfer
• Neural Person Search Machines

Cited by in corpus (16)

• A Strong Baseline and Batch Normalization Neck for Deep Person Re-identification
• Bi-directional Exponential Angular Triplet Loss for RGB-Infrared Person Re-Identification
• Towards Homogeneous Modality Learning and Multi-Granularity Information Exploration for Visible-Infrared Person Re-Identification
• Leaning Compact and Representative Features for Cross-Modality Person Re-Identification
• A heterogeneous branch and multi-level classification network for person re-identification
• Building Computationally Efficient and Well-Generalizing Person Re-Identification Models with Metric Learning
• Compact Network Training for Person ReID
• FPB: Feature Pyramid Branch for Person Re-Identification
• Collaborative Group: Composed Image Retrieval via Consensus Learning from Noisy Annotations
• Grafted network for person re-identification
• AutoLoss: Automated Loss Function Search in Recommendations
• Angular Triplet Loss-based Camera Network for ReID
• Joint Discriminative and Metric Embedding Learning for Person Re-Identification
• On The Distribution of Penultimate Activations of Classification Networks
• Redesigning the classification layer by randomizing the class representation vectors
• Learning Incremental Triplet Margin for Person Re-identification