Research Article
Research on Behavior Recognition Method Based on Machine Learning and Fisher Vector Coding
@INPROCEEDINGS{10.1007/978-3-030-51100-5_12, author={Xing-hua Lu and Zi-yue Yuan and Xiao-hong Lin and Zi-qi Qiu}, title={Research on Behavior Recognition Method Based on Machine Learning and Fisher Vector Coding}, proceedings={Multimedia Technology and Enhanced Learning. Second EAI International Conference, ICMTEL 2020, Leicester, UK, April 10-11, 2020, Proceedings, Part I}, proceedings_a={ICMTEL}, year={2020}, month={7}, keywords={Machine learning Fisher vector coding Behavior recognition}, doi={10.1007/978-3-030-51100-5_12} }- Xing-hua Lu
Zi-yue Yuan
Xiao-hong Lin
Zi-qi Qiu
Year: 2020
Research on Behavior Recognition Method Based on Machine Learning and Fisher Vector Coding
ICMTEL
Springer
DOI: 10.1007/978-3-030-51100-5_12
Abstract
Aiming at the problem that the existing behavior recognition method can not extract the human body interaction area, resulting in low recognition rate, a behavior recognition method based on machine learning and Fisher vector coding is proposed. Constructing artificial neural network based on machine learning, designing the main steps of backward propagation neural network, making the cost function minimum; using the depth continuity of the image to extract the foreground part of the video motion, multiplying with the corresponding 2D video frame to detect the time domain motion Behavior; Solving the dual quadratic programming problem of Fisher support vector machine, obtaining its optimal solution and completing behavior learning; segmenting the current frame image, solving the normal vector to extract the moving target, and completing the behavior recognition method based on machine learning and Fisher vector coding the study. In order to verify the effectiveness of the design method, a comparative experiment was designed. The experimental results show that the average recognition accuracy of the design method is 7.6% higher than the traditional method.
