1 code implementation • 14 Oct 2022 • Van-Anh Nguyen, Khanh Pham Dinh, Long Tung Vuong, Thanh-Toan Do, Quan Hung Tran, Dinh Phung, Trung Le
Our approach departs from the computational process of ViTs with a focus on visualizing the local and global information in input images and the latent feature embeddings at multiple levels.