Second-order Non-local Attention Networks for Person Re-identification
This work addresses the challenge of learning discriminative representations from non-local relationships in person re-identification, which is important for surveillance and security applications, and it is incremental as it builds on existing part-based and attention methods.
The paper tackles the problem of modeling distant or non-local positions in feature maps for person re-identification by proposing a novel attention mechanism based on second-order feature statistics, combined with a generalized DropBlock module. The method achieves results equal to or better than state-of-the-art on datasets such as Market1501, CUHK03, and DukeMTMC-reID.
Recent efforts have shown promising results for person re-identification by designing part-based architectures to allow a neural network to learn discriminative representations from semantically coherent parts. Some efforts use soft attention to reallocate distant outliers to their most similar parts, while others adjust part granularity to incorporate more distant positions for learning the relationships. Others seek to generalize part-based methods by introducing a dropout mechanism on consecutive regions of the feature map to enhance distant region relationships. However, only few prior efforts model the distant or non-local positions of the feature map directly for the person re-ID task. In this paper, we propose a novel attention mechanism to directly model long-range relationships via second-order feature statistics. When combined with a generalized DropBlock module, our method performs equally to or better than state-of-the-art results for mainstream person re-identification datasets, including Market1501, CUHK03, and DukeMTMC-reID.