Low-Resolution Self-Attention for Semantic Segmentation
Semantic segmentation tasks naturally require high-resolution information for pixel-wise segmentation and global context information for class prediction.
BibTex:
Before browse our site, please accept our cookies policy