Fetching the paper…

Vision Transformer with Cross-attention by Temporal Shift for Efficient Action Recognition · Around