A Study on the Learning Based Human Pose Recognition

Human pose recognition is considered a well-known process of estimating the human body pose from a single image or a series of video frames. There exist many applications that can benefit from human pose technology e.g. activity recognition, human tracking, 3D gaming, character animation, clinical analysis of human gait and other HCI applications. Due to its many challenges, such as illumination, occlusion, outdoor environment and clothing, it is considered one of the active areas in computer vision. For the last 15 years, Human pose recognition problem significantly gained interest of many researchers and therefore, many techniques were proposed in order to address the challenges of human pose recognition. In this study, we review the recently progressed work in human pose recognition using computer vision feature extraction and machine learning classification techniques. Accordingly, we identify gaps in existing work and give direction for future work.