Publishing Date

December 30, 2020


Users encounter major problems when they shop online. These problems can be summarized in two main points; choosing suitable size and style of their clothes and returning these items if they find out they don’t fit. Accordingly merchants’ profit is affected severely. Our project aims to help decrease the negative impact of this problem on merchants and let these users find their perfect fitted clothes virtually from their home by taking two images with different positions (front position and side position) or client enter his measurements. After taking these images and processing clients measurements , a 3D model for the user with perfect sizes will be generated.

1.1 Purpose of this document

The purpose of this documentation is to represent a detailed description of our system (Online Fitting Room). Online Fitting Room mainly helps client to fit chosen items virtually. This documentation will present a full description about our system which is a mobile application developed using Flutter and Python. We also provide a fulfilled description about each processing stage, inputs and outputs as well.

1.2 Scope of this document

Reports show that design items are the second generally purchased item among online purchases However, purchasers are facing numerous problems which may prevent them from buying clothes on the web. some of these problems, proposed by GSI1 Business, says that they can’t try on clothes and can’t see their quality before buying them,Online Fitting Room scope is to help brand owners to increase their sales and help clients trying their items online also decreases their visits to stores to avoid COVID-19. Online purchases rose because of the lockdowna lot of countries’ economy was extremely affected because of COVID-19 which spread rapidly, so all countries closed their airports and stopped travelling which is one of the main sources of the country’s income.

1.3 System Overview

First of all, the user is asked to upload two input images with different positions (front and side image). User can also enter his body measurements if known or both of them to increase measurements’ accuracy of the 3D model created and then store it in the database using mobile application. Then, images will go through image processing stage that deployed in heroku which starts with generated 3d model for client afterwards, segment cloth from image using tensorflow to get the mask of the item. Moreover,we train neural network using pytorch to map the segmented cloth into texture after all system generated 3d model for item so user can fit the item on his 3d body model.

1.4 System Scope

1. The user will input two images with different positions; front and side images or user can enter his body measurements if he knows them or user can enter both of them to increase accuracy of body measurements detection using mobile application.

2. The system will take user measurements and user images to generate 3D model.

3. The system will take items from vendor and generate 3D model for each item.

4. Client can choose selected items and see if the selected item fits him or not.