The Virtual Try-On System

Abstract

With the rapid growth of technology people are used to online shopping and online auctions to purchase products. This way of transaction has become a major trend globally with its convenience to customers. However, the main issue of buying clothes online is that inability to try it on before it is delivered which is crucial for customer satisfaction. The objective of the present research is to create a virtual dressing room that realistically reflects the appearance and behaviour of garment. A real time virtual dress up system was built which consists of multiple tasks including extraction of the users’ specific body parts from video stream, alignment of models, detecting and resizing input dress images and dress up using Kinect. Extraction of the users’ specific body parts help to create augmented reality environment by separating the user area from the video stream and imposing it onto the virtual environment in the user interface. The 3D locations of the joints were used for positioning, scaling and rotation in order to align the 3D cloth models with the user. Integration of these techniques, resulted in the present virtual try on system. Finally, system evaluation was done via six independent evaluators (50% female) using a pre and post use questionnaire. In the pre scenario questionnaire 100% agreement was seen regarding the waste of time in queue to dressing room during shopping. In the post scenario questionnaire the system was evaluated in 5 aspects each rated from 1-5 using the average score given by the evaluators. Usefulness of the system was scored 4.39 out of 5 (87.78%), ease of use was rated 4.22 out of 5 (84.32 %), ease of learning 4.61 out of 5 (92.22%) , satisfaction reached 4.02 out of 5 (80.46%) and overall system evaluation was rated 4.22 out of 5 (84.4%). Thereby the present virtual try on system can be regarded as acceptable and satisfactory for the customers while resolving the time factor affecting quality shopping. It can be recommended that the video quality could be improved further in the future.