Person Re-identification Based on Pose-guided Generative Adversarial Network
Guang Han, Chao Yang, Jixin Liu, Ning Sun, Xiaofei Li · Proceedings of the ACM Turing Celebration Conference - China · 2020
Person re-identification (re-id) is a challenging and valuable research topic in the field of computer vision. It needs to match person images with the same identity in multiple camera systems. The change of pose is one of the key factors that affect the network to extract robust features. In order to mitigate the influence of pose variations on person re-id, the paper proposes a person re-id method based on a pose-guided generative adversarial network (PG-GAN), which can be used to learn identity-sensitive and pose-insensitive features. The algorithm is composed of a Siamese convolutional neural network (SCNN) and generative adversarial networks (GANs). SCNN is a symmetric structure with ResNet-50. GANs contain multiple pose discriminators and identity discriminators, as well as incorporate pose loss, which requires appearance of the generated image with same identity to be similar. The proposed method has been well performed on pedestrian reidentification datasets.