SVM-HMM based human behavior recognition

Abstract

In the field of human behavior recognition, a GMM-HMM system shows lower classification performance in short image sequences. The SVMHMM, which has been successfully used in speech recognition, is introduced into behavior recognition in this paper. As one of the discriminate models, SVM is able to use less training samples to distinguish the differences of categories than GMM. Therefore, the part of GMM in the GMM-HMM system is replaced by SVM. The experimental results show that the SVM-HMM system achieves better performance for both short and long image sequences.