1.Hierarchical Transformer means is a variation of the Transformer model that processes data in a structured, multi-level way, unlike standard Transformers that treat input as a flat sequence.Swin is one.

2.A visualization of computed with Shifted windows
Image description

3.
Image description

4.Linear Computational Complexity: Unlike earlier vision transformers that had quadratic complexity, the Swin Transformer has linear complexity with respect to image size, making it more efficient.