survey and comparision of human pose estimation

neural network

In this project human pose estimation based on deep neural network is proposed. The main stream of work in this field has been motivated mainly by the first challenge, the need to search in the large space of all possible articulated poses. This work uses a multi task learning called manifold and Eigen decomposition algorithm which results in high precision pose estimates. Human pose estimation in video is usually conducted by matching 2D image features and retrieving relevant 3D human poses. In the process of retriving, the mapping between images and poses is critical. By using global joint localization and local joint detection, pose estimation can be recovered with better accuracy. The approach has the advantage of reasoning about poses in a holistic fashion so that has a better performance can be achieved compared to the traditional methods. Pose estimation is extracted and classified by different poses such as running, walking, playing ,jogging etc.. Human pose estimation finds its application in image recognition, theft and terrorist activities and video games.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.