Behavior recognition method based on deep residual network

Abstract

The invention discloses a behavior recognition method based on a deep residual network. A spatial network and a time network are respectively constructed by using the deep residual network, and the method comprises a training stage and a test stage: in the training stage, extracting an original frame and an optical flow of a training video, and respectively sending the original frame and the optical flow into the spatial network and the time network for training; in a test stage, extracting an original frame and an optical flow of a test video, respectively sending the original frame and the optical flow to the space and time network models obtained by training, and respectively obtaining a score of each category to which each behavior belongs by each model; and the classification scores of the two models are fused, and a final behavior category is judged through a softmax classifier. According to the method, the features effective to the current behavior can be enhanced according to the importance degree of the feature channel, and the smaller features are suppressed, so that the expression ability of the model to the input data is improved. The method has relatively high behaviorrecognition accuracy, and particularly has relatively good performance in some complex actions and actions which are relatively difficult to recognize.

Description

technical field [0001] The invention belongs to the field of computer technology, especially the field of behavior recognition technology, and relates to a method for recognizing video human behavior, in particular to a behavior recognition method based on a deep residual network (Residual Neural Network, ResNet). Background technique [0002] Video action recognition refers to the use of some algorithms to enable computers to automatically recognize actions in image sequences or videos. Firstly, effective visual feature information is extracted from image sequences and videos, and then an appropriate method is used to represent the information. Finally, a classification model is constructed to learn behaviors and realize correct recognition. [0003] Since the appearance of actions in successive frames in a video is very similar, video action recognition models require temporal reasoning about appearance. In behavior recognition, in addition to behavior appearance, complex...

AI Technical Summary

Problems solved by technology

Although convolutional neural networks have achieved great success in image recognition-based tasks, how to effectively model the temporal evolution of videos with deep networks remains unclear.