JCSE, vol. 16, no. 4, pp.199-210, 2022

DOI: http://dx.doi.org/10.5626/JCSE.2022.16.4.199

Recognition and Classification of Human Actions using 2D Pose Estimation and Machine Learning

Monika Dhiman, Akash Sharma, and Sarbjeet Singh
Computer Science and Engineering, University Institute of Engineering and Technology, Panjab University, Chandigarh, India Computer Science and Engineering, University Institute of Engineering and Technology, Panjab University, Chandigarh, India Computer Science and Engineering, University Institute of Engineering and Technology, Panjab University, Chandigarh, India

Abstract: Recognition and classification of human actions is a fundamental but difficult computer vision task that has been studied by several researchers throughout the world in recent years. Pose estimation is a widely used technology to recognize human actions. It has several applications, especially in the field of computer vision, where it can be used to recognize basic as well as complex human actions. This research provides a novel framework for recognition and classification of human actions which includes five categories - standing, walking, waving, punching and kicking. The dataset used for the recognition and classification purpose is generated using the videos that are recorded by using a smart phone and 2D pose estimation technique has been applied to extract the features from the human body. The ML classifiers have been trained on a custom-built dataset. While all algorithms nearly performed well in classification task, LGBM outperformed the rest in terms of accuracy (98.80 %).

Keyword: Action Classification; Action Recognition; Openpose; Pose Estimation; Machine Learning

Full Paper:   424 Downloads, 887 View