Person re-identification by using a method combining DPM and SVM

Person re-identification refers to pedestrian matching under the surveillance of non-overlapping multi-camera, which is usually used in criminal investigation work and image retrieval. Many metric learning methods have achieved good results, but a single global metric learning is difficult to fit discriminant features. Aiming at the person re-identification under single shot, this paper proposes a new method on the basis of metric methods and classifiers, cutting pedestrians into parts and performing local metric learning on each subset of component, meanwhile, carrying out global metric learning combined with SVM classifier in the entire training set. The experimental results show that the proposed method improves the performance of global metric learning method.