Human Pose Estimation Using Body Part Tracking

Abstract

Human pose estimation is a major computer vision problem that means to detect the spatial area (for example coordinates) of human body joints in unconstrained pictures and images. In other ways we can say it is to predict the location of various human key-points(joints and landmarks) such as elbows, knees, neck, shoulder, hips, chest etc. Estimating human pose is quite challenging problem due to the fact that the human body parts are small and hardly visible parts, occlusions and huge variability in articulations. As strong image processing models, convolutional neural networks (CNNs) comes to the rescue. In this report I included different approaches for human pose estimation, literature survey of major approaches for pose estimation, detailed analysis, showed how we can use activity recognition as a use case using estimating of pose for instance here i used drowning detection as an example, also comparison study of Performance of various deep learning models like mobile net on various Datasets like COCO Dataset, MPII dataset etc.