Hessian-regularized spectral clustering for behavior recognition

In recent years, human behavior recognition has become a hot research issue in computer vision, pattern recognition and other fields. There are a large number of image or video samples with unknown labels in real life, and labeling unknown samples is a time-consuming and laborious task. Therefore, this paper adopts unsupervised learning method to study human behavior recognition, that is, this research propose a Hessian-regularized spectral clustering algorithm and apply it to human behavior recognition. This method uses the Hessian matrix to construct the spectral clustering graph, which can make better use of a large amount of unlabeled information. In order to verify the effectiveness of the improved spectral clustering algorithm, a large number of experiments are conducted on UCF-iphone data set, which is a human behavior data set. The experimental results show that the Hessian-regularized spectral clustering algorithm can effectively improve the accuracy of behavior recognition.