Image-Text Co-Decomposition for Text-Supervised Semantic Segmentation | Read Paper on Bytez