Research Area:  Machine Learning

Abstract:

With the continuous advancement of technology, human behavior recognition, as an important scientific research in the field of computer vision, has important research in many fields such as intelligent surveillance, smart home, virtual reality. In the current complex environment, traditional manual methods have been difficult to meet the requirements of high recognition accuracy and applicability. The introduction of deep learning has brought new development directions for behavior recognition. This article mainly summarizes behavior recognition algorithms based on deep learning. Firstly, the research background and significance of behavior recognition are introduced, and then the traditional learning methods and deep learning methods of behavior recognition are discussed and analyzed respectively, and then the structure of algorithmic models and commonly used public data sets are introduced, and finally, the advantages and disadvantages of the various research directions of human behavior recognition methods based on deep learning are analyzed and some suggestions are given in the future research and expansion directions.

Keywords:  
Behavior Recognition
Deep Learning
Data Set
Structural Model

Author(s) Name:  Yin Chen

Publisher name:  IEEE

DOI:  10.1109/AINIT54228.2021.00027

Volume Information:  

Paper Link:   https://ieeexplore.ieee.org/abstract/document/9724729