DPNeXt: A Lightweight Multi-Scale Feature Fusion Framework for Efficient ViT-Based Multi-Task Dense Prediction figure
AlphaXiv 中文概览(可滚动查看)