Open Access Research article

Research on Human Action Recognition Algorithm Based on Video Image

Le Gao1, Yongjie Huang1, Wanting Zhang1, Tian Yang1* and Jin Hong2

1Wuyi University, China

2The Chinese University of Hong Kong, China

Corresponding Author

Received Date:April 14, 2021;  Published Date: April 28, 2021


Human Action recognition is one of the research hotspot in the field of computer vision and artificial intelligence. It can usually be applied in many fields such as virtual reality, intelligent pension, medical monitoring, child monitoring, public safety and business management. Due to the complexity and variability of human movements, it is difficult and challenging to use computers to accurately identify actions. Therefore, it is of great significance to study the algorithm of motion recognition for video images. In the study of motion recognition algorithm, RGB data are used as input, and Slowfast Network and I3D algorithm models are used for training respectively. Experimental results show that the training accuracy of SlowFast algorithm is 96.66%. The training accuracy of I3D algorithm is as high as 99.16%, and both algorithms can accurately identify human movements. This study confirms that the human motion recognition technology and algorithm will have a wide range of application scenarios in the future. After solving the technical problems of low recognition rate, the recognition algorithm and technology will be widely promoted and applied in various fields.

Keywords: Action Recognition; Slow Fast Algorithm; I3D Algorithm; RGB

Signup for Newsletter
Scroll to Top