Pedestrian re-identification model establishment method based on cross-camera self-distillation and application thereof

• Main Authors: GAO CHANGXIN, CHEN JIYUAN, SANNON
• Format: Patent
• Language: Chinese; English
• Published: 06.06.2023
• Subjects: CALCULATING; COMPUTING; COUNTING; PHYSICS

The invention discloses a pedestrian re-identification model establishment method based on cross-camera self-distillation and application thereof, and belongs to the field of pedestrian re-identification, and the method comprises the steps: extracting the features of a pedestrian image through a feature extraction network, clustering, marking a global pseudo label, dividing the clustering category into a plurality of sub-clusters according to cameras, and marking a local pseudo label; a training set is obtained after labeling is completed; constructing a supervision network comprising a feature extraction network and a plurality of classifiers in one-to-one correspondence with the cameras, and classifying the classifiers according to features to obtain classification probability distribution; training the supervised network by using the training set; the training loss comprises classification loss and distillation loss which respectively represent the difference between the classification probability distribu