Fig. 3From: Learning deep features from body and parts for person re-identification in camera networksIllustration of the a positive and b negative subregion pairsBack to article page