Single Camera Body Tracking for Virtual Fitting Room Application

Online shopping through e-commerce has grown rapidly in the last 4 years and clothing is one of the most in-demand products on e-commerce sites. However, many customers still hesitate to buy their clothing online because of concerns that the product may not fit properly. One technology-based solution to this problem is the Virtual Fitting Room, developed in 2011 using the Kinect. To reach a wider range of customers, Virtual Fitting Room technology will need to be further developed to work on a smartphone. This paper is implementing the Virtual Fitting Room's body tracking system using a single optical camera on a smartphone. Viola-Jones algorithm is used to implement our single-camera body tracking technology through OpenCV and Android Studio. The method is to detect the full body first before tracking the upper body, in both cases using haar feature-based cascades to measure the size of the body based on a model of the human body that follows the golden ratio. The system was successfully implemented accurately with the method of finding mode every five measurements and detections with average error equal to 0.566667 cm.