Study on Deep Learning Models for Human Pose Estimation and its Real Time Application
Jyoti Jangade, Kanojia Sindhuben Babulal
Abstract
In computer vision, human pose estimation details the posture of the person’s body structure that can be Kinematic, Planer, and Volumetric in an image or video. However, pose detection is often critical to be driven by distinct human actions. Thus, this survey report analysis the recent progression of the bottom-up and top-down human pose evaluation models. This survey report focuses on 2D and 3D skeleton-based human pose detection from the captured Red Green Blue(RGB) images. We have condensed the performance of the recent pose recognition, tracking, and detection techniques that utilize pose estimation from colour images as captured and then exhibit room for much more refinement in this domain. In this paper, scrutinize the study of human pose estimation models like 2d and 3d HPE for identify human movements such as running, dancing, sport so on and recent computer vision-based advances. This study has included various methods for detecting in two and three dimensions. This paper summarises the deep learning models for HPE, dataset, and challenges.