Virtual 3D Trial Room

Computer Engineering

Shopping online can be a tedious process as the user has no way to see how the clothes would look on them. This problem can be solved by creating a 3D model to do the same. A common way to acquire such models is with a scanner which is not cost-efficient, numerous researches have been made to capture body shape in a more practical setup. One such practical way would be capturing a video from a single monocular RGB camera. Presenting “Virtual 3D trial room”, an application which uses Convolutional Neural Network (CNN) and OpenCV to produce a 3D model of a person and allows them to try on clothes virtually. The monocular video of a person turning around 360 degrees is provided as input by the user, out of these 8 frames are collected to create the 3D model. Dataset is used to train Machine Learning models to get almost accuracy during the construction of the 3D model. This created 3D model can try on clothes from the digital wardrobe which has clothes which have been converted to 3D models. Using the Multi-Garment Network (MGN) algorithm the look and feel of the cloth for the individual can be estimated. The virtual 3d trial room has a wide range of applications; it can be used with the already existing online clothes shopping website to give users a better shopping experience. It would be extremely helpful during this COVID-19 pandemic to promote social distancing and quarantine; this could also be used later on as well. The process would involve various steps such as separating the person from the background, estimating the joint location, creating the texture of the person, etc.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.