Vision Transformers with Mixed-Resolution Tokenization | IEEE Conference Publication | IEEE Xplore