Pedestrian attribute recognition based on dual self-attention mechanism

Recognizing pedestrian attributes has recently obtained increasing attention due to its great potential in person re-identification, recommendation system, and other applications. Existing methods have achieved good results, but these do not fully utilize region information the correlation between attributes. This paper aims at proposing a robust attribute recognition framework. Specifically, we first propose an end-to-end framework for recognition. Secondly, spatial semantic self-attention mechanism is used key points localization bounding boxes generation. Finally, hierarchical strategy proposed, whole global recognition, relevant regions are local Experimental results on two datasets PETA RAP show that mean accuracy reaches 84.63% 82.70%. The heatmap analysis shows our method can effectively improve Compared with existing methods, it achieve better effect.