Improving Deep Visual Representation for Person Re-identification by Global and Local Image-language Association | Read Paper on Bytez