A CAM-Guided Parameter-Free Attention Network for Person Re-Identification

In this paper, we propose a parameter-free attention mechanism based on class activation mapping (CAM) which is novel compared with most of the existing works that train attention without a supervision signal.Our attention is composed of spatial attention and channel attention in the standard way, which indicates that "where" and "what" is more meaningful, respectively.For Spatial Attention, we use class activation mapping as a supervision signal to guide the generation of it directly in space.Thus our approach to spatial attention can pay more attention to the informative pedestrian parts of the scene and reduce background interference.For Channel Attention, the importance of each channel is obtained by the similarity between the aforementioned spatial attention and the feature map of each channel.In this manner, our channel attention is indirectly guided by CAM.In addition, our attention is parameter-free, which reduces the risk of overfitting.Finally, we conduct extensive evaluations on three popular benchmark datasets including Market1501, DukeMTMC-reID, and MSMT17, demonstrating the effectiveness of our approach on discriminative person representations.