Clothing Item Similarity

This project aims to find similar clothing items based on their descriptions using machine learning techniques. It collects and preprocesses data, measures similarity, and returns ranked results.

Problem Statement

The problem is to develop a system that can find similar clothing items based on their descriptions. The system should collect data from multiple e-commerce websites, preprocess the text data, extract useful features, compute similarity between the input text and the database, and return a ranked list of similar item URLs.

Solution

The solution involves the following steps:

1. Collect and preprocess data:

   • Web scrape multiple e-commercelike amazon , flipkart and nykaa fashion websites to gather clothing(Mens : Shirts,T-shirts , jeans Womens : One Piece , Tops) item descriptions and their corresponding URLs.
   • To keep the datasets balance we have gathered equal amaount of entries from the above stated e-commerce websites.
   • Preprocess the text data by cleaning it, removing special characters, lowercasing, etc., and apply text normalization techniques like stemming or lemmatization.

2. Measure similarity:

   • Extracted useful features from the text descriptions using techniques like TF-IDF .
   • Implement a method to compute the similarity between the input text and the texts in the database using cosine similarity.

3. Return ranked results:

   • Designed a function that accepts a text string, extracts its features, computes similarities with the items in the database, ranks them based on similarity, and returns the URLs of the top-N most similar items.

4. Deploy the function:

   • Containerize the application using Docker.
   • Deploy the function on AWS EC2 (Although GCP was recommened due to technical isssues i was not able to access my free trail account but in future we will deploy of function to GCP Function).
   • The function is accepting a JSON payload with the input text and return a JSON response with the ranked list of similar item URLs.

video_link : https://youtu.be/EUFEDruCOvM

I usally love feebacks and suggestion Contact Us:- [email protected] (Hope this repo found it usefull to you then please give a star to this repo).

If you want to contribute to this project feel free to contact me or just genrate a issue . Note : There is a lot of potential in this project in future we will adding more data and other properties to suggest more accurately .

Thanks ❤️❤️❤️ shashank Bhardwaj(the_ghost) ❤️❤️❤️❤️