Distilling CLIP with Dual Guidance for Learning Discriminative Human Body Shape Representation | Read Paper on Bytez