PF-ViT: Parallel and Fast Vision Transformer for Offline Handwritten Chinese Character Recognition

Figure 2

(a) Transformer encoder module; (b) multihead self-attention head; (c) self-attention head.