Efficient Learnable Collaborative Attention for Single Image Super-Resolution

arXiv:2404.04922v1 Announce Type: new
Abstract: Non-Local Attention (NLA) is a powerful technique for capturing long-range feature correlations in deep single image super-resolution (SR). However, NLA suffers from high computational complexity and memory consumption, as it requires aggregating all non-local feature information for each query response and recalculating the similarity weight distribution for different abstraction levels of features. To address these challenges, we propose a novel Learnable Collaborative Attention (LCoA) that introduces inductive bias into non-local modeling. Our LCoA consists of …

abstract arxiv attention collaborative complexity computational consumption correlations cs.ai cs.cv distribution feature however image information local attention memory memory consumption query resolution type