Marker-Free Computer Vision for Human Motion Analysis: A Review.

Vision based human motion analysis encapsulates various tasks ranging from gesture recognition, pose detection, tracking, etc. to complex behavior analysis crucial in the realms of sports analytics, gait analysis in athletes, etc. With the advancement in deep learning, marker-free human motion analysis methods have been explored for possible wider applications such as smart surveillance, health monitoring, biokinematics, virtual reality, etc. This paper provides a summary of marker-free human motion analysis in the last decade. The main emphasis is on three main components (a) target application, (b) complexity and (c) baseline algorithms employed in general human motion analysis systems. It presents a novel hierarchical categorization of computer vision frameworks and guides the readers to relevant literature based on their respective challenging scenarios. We also discuss real-world applications and examine the state-of-the-art with respect to the datasets and quantitative results.