A median filter detection algorithm based on multi-residual learning and attention fusion is proposed for small size JPEG compressed images carrying less effective information and less obvious median fi lter traces.In this algorithm,multiple high-pass filters are combined with the attention module to obtain multi-residual feature maps wi th weights as the input of the feature extraction layer, and the feature extraction layer adopts the form of grouped convolution to perfo rm multi-scale feature extraction on the input multi- residual feature maps and fuse the feature information with different scales,whil e the dense connection is used,and the input of each convolution layer comes from the output sum of all previous convolution layers. The experimental results show that the proposed algorithm in this paper has higher detection accuracy than existing algorithms f or median filter detection of small JPEG compressed images,and can detect and locate local tampered regions more effectively.