Generative Adversarial Network (GAN) Based Human Pose Recognition

  • Gnanavel R, Jeba Singh P,Hari S, Krishna Viswanathan S

Abstract

Pose estimation is the technique of determining human poses and gestures. This approach does not aim to figure out who is in the image, but rather focuses on the crucial body joints for estimating the person’s pose. A problem that holds great significance in the field of Computer Vision, real time analysis of pose estimation could be extremely advantageous. From gaming to medicine, human pose estimation finds its application in a wide range of sectors. This project aspires to automate the process of pose estimation, with the help of Generative Adversarial Network. General adversarial network helps to predict the key points more accurately, Which improves the performance of human pose estimation.The Existing system is based on covolutional neural networks, which cannot detect the footprints and the existing research needs high end systems to operate. All these problems are solved in the proposed work. Genrative Adversarial Network improves the performance of Human Pose Estimation. Open Pose Estimation using GAN provides more accurate results than convolutional neural networks.

Published
2020-04-04
How to Cite
Gnanavel R, Jeba Singh P,Hari S, Krishna Viswanathan S. (2020). Generative Adversarial Network (GAN) Based Human Pose Recognition. International Journal of Advanced Science and Technology, 29(04), 1805 - 1809. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/7896