Pixel-level Semantic Correspondence through Layout-aware Representation Learning and Multi-scale Matching Integration | Read Paper on Bytez