Motion recognition method based on three-dimensional convolution depth neural network and depth video

Abstract

The invention discloses a motion recognition method based on a three-dimensional convolution depth neural network and depth video. In the invention, depth video is used as the object of study, a three-dimensional convolution depth neural network is constructed to automatically learn temporal and spatial characteristics of human body behaviors, and a Softmax classifier is used for the classification and recognition of human body behaviors. The proposed method by the invention can effectively extract the potential characteristics of human body behaviors, and can not only obtain good recognition results on an MSR-Action3D data set, but also obtain good recognition results on a UTKinect-Action3D data set.

Description

technical field [0001] The invention relates to the field of action recognition, in particular to an action recognition method based on a three-dimensional convolutional deep neural network and deep video. Background technique [0002] As a popular technology in video analysis, human behavior recognition has gradually begun to be applied in daily life, such as abnormal event detection in automatic surveillance, video retrieval, human-machine interface, etc. Traditional human action recognition includes three steps: feature extraction, feature representation and recognition classification. First, artificial features are extracted from video sequences. Secondly, some techniques such as transformation and clustering are used to construct more discriminative descriptor features from the extracted features. Finally, a classifier is used to classify and identify the descriptor features. The behavior recognition method based on feature extraction has achieved very promising rese...

AI Technical Summary

Problems solved by technology

However, the research progress of behavior recognition methods based on artificial features has been slow in recent years, mainly because, first, in order to effectively retain behavior information, the extracted feature dimensions are getting higher and higher, and the calculation overhead is too large, so it is difficult to achieve real-time performance. ; Second, the artificially designed features are adjusted for a specific data set, and it is difficult to generalize to other data sets; Third, the steps of traditional behavior recognition methods are isolated, and the classification results cannot be automatically fed back to the feature extraction and description

In short, the traditional human behavior recognition method based on artificially designed features involves many links, has the disadvantages of high time consumption and difficulty in overall algorithm optimization

Images