RepViT：从ViT视角重新审视移动CNN

摘要

https://arxiv.org/pdf/2307.09283.pdf
近年来，与轻量级卷积神经网络(CNN)相比，轻量级视觉Transformers(Vi