5. 可视化分析:图 3 的可视化结果显示出了一个惊奇的现象——简单的 ToMe 就可以将相同的实例对应的 Token 合并在一起,无论是前景还是背景。
▲ 图3. ViT-H MAE ImageNet-1k 在训练过程中应用 ToMe 的融合结果
参考文献
[1] Xiaoyi Dong, Jianmin Bao, Dongdong Chen, Weiming Zhang, Nenghai Yu, Lu Yuan, Dong Chen, and Baining Guo. Cswin transformer: A general vision transformer backbone with cross-shaped windows. In CVPR, 2022.[2] Haoqi Fan, Bo Xiong, Karttikeya Mangalam, Yanghao Li, Zhicheng Yan, Jitendra Malik, and Christoph Feichtenhofer. Multiscale vision transformers. In ICCV, 2021.[3] Benjamin Graham, Alaaeldin El-Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Herv ́e J ́egou, and Matthijs Douze. Levit: a vision transformer in convnet's clothing for faster inference. In ICCV, 2021.[4] Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll ́ar, and Ross Girshick. Masked autoencoders are scalable vision learners. In CVPR, 2022.[5] Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. An image is worth 16x16 words: Transformers for image recognition at scale. In ICLR, 2020.[6] Dmitrii Marin, Jen-Hao Rick Chang, Anurag Ranjan, Anish Prabhu, Mohammad Rastegari, and Oncel Tuzel. Token pooling in vision transformers. arXiv:2110.03860 [cs.CV], 2021.