Weidong Zhang

PathViT: an efficient scalable multi-scale conv-attentional vision transformer for few-shot image classification

This paper redesigns the internal space structure of Transformers and proposes a novel Factorized Multi-head Self-attention with Enhanced-side (FMSA), and constructs a fusion block based on multi-scale convolution and FMSA, which can fully exploit the advantages of local inductive biases in ConvNets and global context...