Semantic segmentation tasks naturally require high-resolution information for pixel-wise segmentation and global context information for class prediction. While existing vision transformers demonstrate promising performance, they often utilize high-resolution context modeling, resulting in a computational bottleneck. In this work, we challenge conventional wisdom and introduce the Low-Resolution Self-Attention (LRSA) mechanism to capture global context at a significantly reduced computational cost, i.e., FLOPs. Our approach involves computing self-attention in a fixed low-resolution space regardless of the input image's resolution, with additional 3x3 depth-wise convolutions to capture fine details in the high-resolution space. We demonstrate the effectiveness of our LRSA approach by building the LRFormer, a vision transformer with an encoder-decoder structure. Extensive experiments on the ADE20K, COCO-Stuff, and Cityscapes datasets demonstrate that LRFormer outperforms state-of-the-art models. Code is available at https://github.com/yuhuan-wu/LRFormer.
@article{arxiv.2310.05026,
title = {Low-Resolution Self-Attention for Semantic Segmentation},
author = {Yu-Huan Wu and Shi-Chen Zhang and Yun Liu and Le Zhang and Xin Zhan and Daquan Zhou and Jiashi Feng and Ming-Ming Cheng and Liangli Zhen},
journal= {arXiv preprint arXiv:2310.05026},
year = {2025}
}
Comments
Accepted by IEEE TPAMI; 14 pages, 6 figures, 14 tables