Learning Compact Appearance Representation for Video-based Person Re-Identification | Read Paper on Bytez