Multi-scale Deep Learning Architectures for Person Re-identification

By Xuelin Qian, Yanwei Fu, Yu-Gang Jiang, Tao A. Xiang, Xiangyang Xue. IEEE International Conference on Computer Vision, 2017

305 citations recorded by the source corpus.

Person Re-identification (re-id) aims to match people across non-overlapping camera views in a public space. It is a challenging problem because many people captured in surveillance videos wear similar clothes. Consequently, the differences in their appearance are often subtle and only detectable at the right location and scales. Existing re-id models, particularly the recently proposed deep learning based ones match people at a single scale. In contrast, in this paper, a novel multi-scale deep learning model is proposed. Our model is able to learn deep discriminative feature representations at different scales and automatically determine the most suitable scales for matching. The importance of different spatial locations for extracting discriminative features is also learned explicitly. Experiments are carried out to demonstrate that the proposed model outperforms the state-of-the art on a number of benchmarks.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/01381e2faeb26489ae9c24c84c42df33901cfe61

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Yi Yang
2. Tao A. Xiang
3. Ling Shao
4. Dacheng Tao
5. Luc Van Gool
6. Qi Tian
7. Liang Zheng
8. Xiaogang Wang

Citations in the corpus

• Pose-Normalized Image Generation for Person Re-identification (2017, methodology): “Deep re-id models Most recently proposed re-id models employ a DNN to learn discriminative view-invariant features [25, 2, 9, 51, 34, 40, 63, 47].”
• Harmonious Attention Network for Person Re-identification (2018, background): “Most existing person re-id methods focus on supervised learning of identity-discriminative information, including ranking by pairwise constraints [23, 38], discriminative distance metric learning [13, 45, 40, 20, 41, 4], and deep learning [24, 18, 3, 39, 35, 19].”
• Pedestrian Alignment Network for Large-scale Person Re-Identification (2017, background): “[42] deepen the network and use filters of smaller size.”
• Camera Style Adaptation for Person Re-identification (2017, unknown): “Many deep learning methods [33, 30, 29, 3, 20] have been proposed in person re-ID.”
• Dual-path Convolutional Image-Text Embeddings with Instance Loss (2017, background): “In the experiment, despite of the limited training data, the learned model has a good generalization ability on the validation set and test set, which accords with existing experience [56], [57].”
• Deep Representation Learning With Part Loss for Person Re-Identification (2017, background): “, DeepReID [19], LSTM Siamese [36], Gated Siamese [21], and MuDeep [64].”
• Pose-Guided Feature Alignment for Occluded Person Re-Identification (2019, background): “Recent years, deep learning methods [39, 34, 13, 27, 3, 25, 7, 6, 36, 21] for person re-id show significant superiority on retrieval accuracy.”
• Omni-Scale Feature Learning for Person Re-Identification (2019, background): “Different to [2], the network in [30] is built on Inception [44, 45], where multiple streams were originally designed for low computational cost with handcrafted mixture of convolution and pooling layers.”
• Deep Learning for Person Re-Identification: A Survey and Outlook (2020, background): “Qian et al. [77] develop a multi-scale deep representation learning model to capture discriminative cues at different scales and adaptively mine the appropriate scale for person retrieval.”