Vision-Based Human Pose Estimation for Home Fitness

• Main Authors: Lee,Chao-Yu, 李超宇
• Other Authors: 葉經緯
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi/login?o=dnclcdr&s=id=%22106CCU00442005%22.&searchmode=basic

碩士 === 國立中正大學 === 電機工程研究所 === 106 === Human posture identification remains to be one of the keys for technologies to prevail in human lives, especially for those gadgets that need to interact with users. This thesis explores using only images taken from a typical webcam to identify human posture in real time. In particular, we adapt a state-of-the-art, image-based human posture estimation algorithm for personal exercise application, and accordingly propose several improvements. Experimental results shows that our proposition is able to improve the operation speed by four times and to increase the accuracy of identification from 70% to 90%.