Skip to content

HemanthPaila/FitPic

Repository files navigation

👕 FitPic – AI-Powered Personalized Fashion Recommendation System

📌 Project Overview

FitPic is a full-stack, AI-driven web application that delivers personalized clothing recommendations using computer vision and machine learning.

During authentication, users upload or capture an image. The system analyzes visual attributes such as:

  • Skin tone
  • Face shape
  • Hair color
  • Eye type

Using these extracted features, FitPic intelligently generates outfit combinations from multiple fashion brands and assigns a compatibility score (e.g., 80% match). The platform then recommends the most suitable outfits tailored to each user.

FitPic aims to simplify online fashion decisions, improve styling confidence, and reduce product return rates.


🚀 Problem Statement

Online shoppers often struggle with:

  • Choosing colors that suit their skin tone
  • Identifying styles that match their facial structure
  • Pairing clothing items effectively

Most e-commerce platforms provide generic recommendations without personalization.

👉 FitPic solves this by delivering data-driven, personalized outfit recommendations based on individual visual attributes.


🧠 How It Works

1. 🔐 User Authentication

  • User registers or logs in
  • Image is uploaded during authentication

2. 📷 Attribute Extraction (AI Model)

Computer vision models analyze:

  • Skin tone
  • Face shape
  • Hair color
  • Eye type

3. 🛍️ Product Collection

  • Clothing data is collected from multiple brand sources
  • Structured into datasets for processing

4. 🎯 Pair Generation Algorithm

  • Clothing items (shirts, pants, etc.) are automatically paired

  • Compatibility score is calculated using:

    • User attributes
    • Style rules
    • Color theory

5. 📊 Personalized Recommendations

  • Top outfit combinations are displayed
  • Each includes a match percentage score

6. 💳 Product Details & Checkout

  • Product preview in UI
  • Secure payment gateway integration

🖥️ Features

  • 🔐 User Authentication
  • 📷 AI-Based Image Analysis
  • 🎯 Attribute-Based Outfit Matching
  • 📊 Compatibility Score System
  • 🛍️ Multi-Brand Outfit Pairing
  • 📱 Fully Responsive UI (Mobile / Tablet / Desktop)
  • 💳 Secure Payment Integration
  • ⚡ Fast API-based Backend

🏗️ Tech Stack

🔹 Frontend

  • React.js
  • HTML5 / CSS3
  • Responsive UI Design

🔹 Backend

  • FastAPI (Python)
  • REST API Architecture

🔹 Machine Learning

  • TensorFlow (Fit + Try-On Models)
  • OpenCV (Image Processing)

🔹 Database

  • MongoDB

🔹 DevOps & Tools

  • Docker & Docker Compose
  • GitHub (Version Control)

📁 Project Structure

fitpic/
│
├── backend/
│   ├── app/
│   │   ├── main.py
│   │   ├── api/routes/
│   │   ├── services/
│   │   ├── models/
│   │   ├── schemas/
│   │   ├── utils/
│   │   ├── db/
│   │
│   ├── requirements.txt
│   ├── Dockerfile
│
├── frontend/
│   ├── src/
│   │   ├── components/
│   │   ├── api/
│   │   ├── App.js
│
├── ml/
│   ├── train_fit.py
│   ├── train_tryon.py
│
├── docker-compose.yml
└── README.md

⚙️ Installation & Setup

🔹 Prerequisites

  • Docker & Docker Compose
  • Node.js & npm

🔹 Run the Project

docker-compose up --build

🤖 Machine Learning Components

🔹 Fit Recommendation Model

  • Input: Body + attribute features

  • Output:

    • Fit score
    • Recommended size

🔹 Virtual Try-On Model (Basic)

  • Input:

    • User image
    • Clothing image
  • Output:

    • Generated try-on image

⚠️ Note: Current try-on model is a baseline implementation. Production-grade systems should use advanced models like diffusion-based try-on.


📊 Future Enhancements

  • 👕 Virtual Try-On (Advanced / Diffusion Models)
  • 📏 Body Measurement Estimation
  • 🧠 AI Chat Styling Assistant
  • 📈 Style History Tracking
  • 🌦️ Seasonal Trend Adaptation
  • 🧍 3D Avatar Integration (Three.js)

🎯 Project Vision

FitPic aims to become a smart fashion assistant that combines AI with styling intelligence to deliver:

  • Personalized outfit recommendations
  • Improved shopping confidence
  • Reduced product return rates

⚠️ Current Limitations

  • Try-on model is not production-level
  • Requires real-world datasets for accuracy
  • Styling rules are basic (can be enhanced with ML + fashion datasets)

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors