Workflow from Scientific Research

The network structure of two successive Swin Transformer blocks involved in the entire network architecture. The Swin Transformer architecture includes a classification output head with a global average pooling layer followed by a linear layer, and the hidden layer in the first stage has 128 channels. Input images are divided into non-overlapping windows via patch partition, enabling self-attention within these windows to capture local structures and patterns effectively.

DOI