Multiview human action recognition system based on OpenPose and KNN classifier

• Main Authors: Malik, Najeeb Ur Rehman, Abu Bakar, Syed Abdul Rahman, Sheikh, Usman Ullah
• Format: Book Section
• Published: Springer Science and Business Media Deutschland GmbH 2022
• Subjects: QA75 Electronic computers. Computer science; TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://eprints.utm.my/id/eprint/100671/; http://dx.doi.org/10.1007/978-981-16-8129-5_136

Human action recognition is one of the trending research topics in the field of computer vision. Human-computer interaction and video monitoring are broad applications that aid in the understanding of human action in a video. The problem with action recognition algorithms such as 3D CNN, Two-stream network, and CNN-LSTM is that they have highly complex models including a lot of parameters resulting in difficulty while training them. Such models require high configuration machines for real-time human action recognition. Therefore, present research proposes the use of 2D skeleton features along with a KNN classifier based HAR system to overcome the aforementioned problems of complexity and response time.