Problem Statement: Finding Visually Similar Garments for an Input Garment

Garments in the fashion domain can be of multiple shapes, sizes, and colors. Finding garments similar to each other is an important feature used by e-commerce websites to show recommendations to its users. We would like to find visually similar garments for any input garment from within a given dataset of garment images.

Demo

demo.mp4

Architecture

Dataset

https://drive.google.com/file/d/1OCvfi5L_znC3xGGyH_hXEYEKSGcRleHU/view?usp=sharing

How to use this repository

• Step1: git clone https://github.com/sayansaha934/Garments-Recommendation-System.git
• Step2: Change directory to Garments-Recommendation-System
• Step3: Create a directory database inside static
• Step4: Download the dataset and store all images in static/database
• Step5: Create a virtual env conda create -n Garments-Recommendation-System python==3.6.9
• Step6: conda activate Garments-Recommendation-System
• Step7: Install requirements pip install -r requirements.txt
• Step8: Run store_features.py (This step extracts features from all images and stores in features.pkl file)
• Step9: Run app.py

Future works

All images and its corressponding feature vector will be stored in a database.