The Journal of China Universities of Posts and Telecommunications ›› 2022, Vol. 29 ›› Issue (5): 21-29.doi: 10.19682/j.cnki.1005-8885.2022.0007

Special Issue: Special Topic on Artificial Intelligence of Things

Previous Articles     Next Articles

Saliency guided self-attention network for pedestrian attribute recognition in surveillance scenarios

Li Na1,2, Wu Yangyang1, Liu Ying1,2, Li Daxiang1,2, Gao Jiale1   

  • Received:2021-05-12 Revised:2021-09-16 Online:2022-10-31 Published:2022-10-28


Pedestrian attribute recognition is often considered as a multi-label image classification task. In order to make full use of attribute-related location information, a saliency guided sel-attention network ( SGSA-Net) was proposed to weakly supervise attribute localization,without annotations of attribute-related regions. Saliency priors were integrated into the spatial attention module ( SAM ). Meanwhile, channel-wise attention and spatial attention were introduced into the network. Moreover, a weighted binary cross-entropy loss ( WCEL) function was employed to handle the imbalance of training data. Extensive experiments on richly annotated pedestrian ( RAP) and pedestrian attribute ( PETA) datasets demonstrated that SGSA-Net outperformed other state-of-the-art methods.

Key words: pedestrian attribute recognition|saliency detection|self-attention mechanism