Human body action recognition method and system based on deep learning

Abstract

The invention discloses a human body action recognition method based on deep learning. The method comprises the steps of obtaining two continuous frames of images in a video sequence, inputting the two continuous frames of images into a trained human body action recognition model to obtain a human body action recognition result, wherein the human body action recognition model is generated throughthe following steps: obtaining two continuous frames of images in the video sequence in the data set, extracting an optical flow image from two acquired continuous frames of images by utilizing an optical flow extraction method; and repeatedly executing the process on all the remaining frames in the video sequence, averagely dividing the video sequence and the optical flow image sequence into T segments, extracting a single-frame image from each segment of the video sequence, and extracting continuous L frames of optical flow images from each segment of the optical flow image sequence. According to the method, the technical problems that the extracted features are not beneficial to classification and the recognition accuracy is low due to the fact that the depth of the extracted network issmall and the time sequence relation between the features is not considered in an existing human body action recognition method can be solved.

Description

technical field [0001] The invention belongs to the technical field of deep learning, and more specifically, relates to a human action recognition method and system based on deep learning. Background technique [0002] The traditional human motion recognition is to add acquisition devices such as biosensors or mechanical sensors to the human body. It is a contact motion detection method, which will bring disgust or fatigue to people. With the development of technology, this recognition mode has been gradually replaced by image-based recognition methods. [0003] The introduction of deep learning has made breakthroughs in machine learning, and has also brought a new development direction for human action recognition. Different from traditional recognition methods, deep learning can automatically learn high-level features from low-level features, which solves the problem that feature selection is too dependent on the task itself and the adjustment process takes a long time. ...

AI Technical Summary

Problems solved by technology

[0006] In view of the above defects or improvement needs of the prior art, the present invention provides a human action recognition method and system based on deep learning. The timing relationship between features will lead to technical problems that the extracted features are not conducive to classification and low recognition accuracy

Images