FashionLLM - Custom Fashion Generator Project

Have you ever wanted to effortlessly embody the style of your favorite celebrities or iconic personalities?

FashionLLM is a unique AI-powered project that generates custom fashion styles inspired by the aesthetics of famous individuals. This repository encompasses a full-stack application that combines cutting-edge AI models, web scraping techniques, and modern web development tools to provide users with a personalized fashion experience. Whether you're a fashion enthusiast or an AI aficionado, FashionLLM offers an engaging platform that seamlessly integrates technology and style

Introduction

FashionLLM leverages the power of Language Models and AI to generate fashion suggestions based on the styles of various iconic figures. By using AI models and web scraping, the project gathers data from fashion websites, processes it, and recommends clothing items that align with a selected persona's style.

See How it Looks

See FashionLLM in action!

Click to watch MVP video

The Everyday Solution

In the hustle and bustle of our day-to-day lives, finding the perfect clothing style that resonates with our personality can be a challenging task. Not everyone has the time to delve into fashion trends, fabric choices, and ensemble coordination. This is where FashionLLM steps in to offer a seamless and innovative solution.

Fashion for Every Occasion: Whether you're preparing for a casual outing, a formal event, or even a themed party, FashionLLM enables you to effortlessly select a style that suits the occasion.

Personalized Style Recommendations: Parents, for instance, can find great relief in using FashionLLM to select clothing for their children. Picture a scenario where a child has just started learning football and is a devoted fan of Leo Messi. With FashionLLM, parents can easily browse and order clothes inspired by Leo Messi's distinctive style, ensuring their child feels connected to their idol while staying comfortable.

Time-Efficient and Hassle-Free: Say goodbye to hours of scrolling through online stores and wondering which outfit matches your vision. FashionLLM simplifies the process by providing curated fashion recommendations that align with your chosen persona's style.

Empowering Self-Expression: Fashion is a powerful form of self-expression. FashionLLM empowers individuals to express themselves through clothing, enabling them to experiment with new styles and explore different fashion directions.

Introduction
Features
Project Structure
Technologies Used
Getting Started
Usage
Contributing
License

Features

AI-powered Fashion Generation: Generate custom fashion recommendations inspired by the style of any famous person.
Web Scraping: Scrape fashion data from popular websites like Asos and Shein to provide a wide range of clothing options.
User Profiles: Allow users to create profiles, save favorite looks, and receive personalized fashion suggestions.
MongoDB Integration: Store scraped fashion data and user preferences using MongoDB.
User Management: Manage user data and authentication using a MySQL database.
Caching with Redis: Improve application performance through caching using Redis.
RESTful API: Backend API built using FastAPI to handle data requests and responses efficiently.
Modern Frontend: Develop the web application frontend using React and TypeScript for a dynamic and responsive user interface.

Project Structure

Backend: Contains the backend implementation of the project, including AI models, API routes, and database integration.
Frontend: Contains the frontend implementation using React and TypeScript, providing users with an interactive interface
Scraper : Contains the scraping functionality for various fashion websites such as Asos and Shein (more to come soon)
notebooks: Houses utility scripts for tasks like data scraping, model training, and data science operations.

Technologies Used

Backend Server:

Python
OpenAI GPT-3 and LangChain
FastAPI (and Pydantic)
Docker
MongoDB
MySQL
Redis

Frontend Client:

React
TypeScript
CSS, Bootstrap, Material ui
Redux

Usage

Create a user profile on the FashionLLM platform.
Explore the available famous personas and select a style you'd like to adopt.
Receive fashion recommendations based on the chosen persona's style.
Save favorite looks, explore more styles, and fine-tune your preferences over time.

What the Future Holds

I have exciting plans for the future of FashionLLM:

Full Web Application Launch: Soon, I will launch the complete web application, allowing users to experience the full range of features and functionalities.
Expanded Web Fashion Website Scrapers: Im actively working on adding more web fashion website scrapers to provide users with an even broader selection of fashion items to choose from.
Enhanced Custom Styling with AI Models: Im developing AI models to recommend not only clothing items based on celebrites fashion but also fashion design products based on user-defined "themes," allowing for more personalized and creative styling.

Frontend Installation

Make sure Nodejs and NPM are installed in your machine!

Nodejs and NPM installation guide: Nodejs and NPM installation

To set up the frontend of the FashionLLM application, follow these steps:

Clone this repository if you haven't already:

git clone https://github.com/avivohayon/FashionLLM.git

Navigate to the frontend directory:
```
cd FashionLLM/Frontend/fashion-app
```
Install the required frontend dependencies:
```
npm install
```
inside fashion-app run:
```
 npm run dev 
```

and lunch the client server

important! make sure you're using the correct port for the frontend

Local: http://localhost:5173/

Backend Installation

Make sure python10 or above is installed in your machine!

Python10 installation guide: Python10 installation

To set up the backend of the FashionLLM application, follow these steps:

Make sure you have Python and pip installed on your system. You can download them from here.

Clone this repository if you haven't already:

git clone https://github.com/avivohayon/FashionLLM.git

Navigate to the backend directory:
```
cd FashionLLM
```

Create a virtual environment (optional but recommended):

python -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate

Install the required Python packages:
```
pip install -r requirements.txt
```
Set PYTHONPATH:

open your cmd in directory where you clone this project for example: "C:\Desktop\study\test\testFashionLLM\FashionLLM"

run the following command for setting the python path:
```
set PYTHONPATH=<the_path_of_the_clone_project> #in my example: "set PYTHONPATH=C:\Desktop\study\test\testFashionLLM\FashionLLM"
```

Now you can run the backend server by exectue the command:

cd <the_path_of_the_clone_project>/FashionLLM/Backend # in my example: "cd C:\Desktop\study\test\testFashionLLM\FashionLLM\Backend
python main.py # run backend server

Set Up Docker Containers:

Install Docker Desktop on your system: Docker Desktop Installation

Pull and run the Redis container:

docker run -d --name fashion-redis -p 6379:6379 redis

Pull and run the MySQL container:(notice the port numbers 3307:3306 )

docker run -d --name fashion-mysql -e MYSQL_ROOT_PASSWORD=your_root_password -e MYSQL_DATABASE=your_database_name -p 3307:3306 mysql:latest

Pull and run MongoDB container:

docker run -d -p 27017:27017 --name mongodb-fashion-container mongo:latest

.env file and needed info

.env variables needed :

HASH_ALGORITHM
DATABASE_URL
OPENAI_API_KEY
CRYPT_SECRET_KEY

In FashionLLM/Backend folder create .env file with these vars:

    HASH_ALGORITHM=HS256
    DATABASE_URL=mysql+pymysql://root:<your_root_password>@localhost:3307/UsersDB
    OPENAI_API_KEY=<your_openai_api_key>

in your terminal write:

   openssl rand -hex32

copy the printed value and paste it to CRYPT_SECRET_KEY

CRYPT_SECRET_KEY=<hashed_value>

if you have problem init the mongoDb database do this

cd FashionLLM\Backend\database
run python config_structure.py

techinAI/FashionLLM