SphereReID: Deep Hypersphere Manifold Embedding for Person Re-Identification

TL;DR

This paper introduces SphereReID, a deep learning model that uses a hypersphere manifold embedding with Sphere Softmax to improve person re-identification accuracy across multiple challenging datasets.

Contribution

The paper proposes SphereReID with Sphere Softmax and a balanced sampling strategy, enabling end-to-end training that surpasses state-of-the-art methods without re-ranking.

Findings

Achieves 94.4% rank-1 accuracy on Market-1501

Achieves 83.9% rank-1 accuracy on DukeMTMC-reID

Outperforms existing methods on four datasets

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.