An ICT Based Solution for Virtual Garment Fitting in for Online Marketplace

Abstract

Abstract There is a considerable online shopping rate corresponding to apparels, yet high return rate is visible in online apparel purchases. Size mismatch is a major reason behind the issue of returning apparels purchased online. In the context of virtual fitting, there are many 3D applications to check the fitness of a particular garment to a specific user. The applicability of these 3D solutions for the online marketplace is comparatively low since they require high user intervention and high computational power. Hence 3D solutions cannot be easily used to obtain a real time result for the online marketplace, even with their ability to provide accurate outcomes regarding the fitness of a garment. This research presents a lightweight solution which supports the decision making process of an end user in order to select properly fitted cloths by considering human body properties and garment properties. A fitting model is proposed by incorporating the 2D patterns of the human body and the finished garment and is successfully demonstrated in this research. The fitting model has the capability of identifying the fitness of seven predefined areas of the body namely shoulder length, chest, stomach, hip, sleeve length, sleeve tightness and shirt length. The scope of the research has been limited to the generation of a fitting model for short sleeve men’s shirts due to the limited time frame. An experiment has conducted to discover the best method for generating 2D block pattern from a finished shirt which identified the standard error values associated in the conversion between finished shirt and 2D block pattern. These error values were generalized and used for generating the fitting model. User preference regarding fitness priority and distance ease values have been incorporated to the fitting model. A prototype system was implemented based on the proposed methodology and satisfactory outcomes have been obtained. A survey has been carried out to identify the aspects related to segmentation of a shirt based on prioritizing fitness areas. Qualitative approaches were used to evaluate the implemented model and it was found that implemented system performs with an accuracy of 81.2%. A user evaluation survey was conducted to find out the effectiveness of the proposed visualization techniques and 76% of the participants agreed to the fact that the current visualization system is suitable to determine the fitness of the garment. The research methodology is scalable to other garment items and other genders though the research was conducted only for short sleeve men’s shirts.