💇 HairStyle AI

AI-powered face shape analysis & personalised hairstyle + beard style recommendations
Built with Python · Flask · MediaPipe · MongoDB · Gemini · Replicate

---

🌟 Overview

HairStyle AI is a web application that analyses a user's face shape from a photo and instantly recommends the most flattering hairstyles and beard styles. It uses Google's MediaPipe to detect 468 facial landmarks, classifies the face shape, and provides curated style recommendations — complete with an AR Try-On feature powered by Replicate's image generation API.

---

✨ Features

Feature	Description
📸 Face Shape Analysis	Upload a photo — AI detects your face shape (Oval, Round, Square, Heart, Diamond, Oblong)
💇 Hairstyle Recommendations	Get personalised hairstyle suggestions tailored to your face shape
🧔 Beard Style Recommendations	Beard style recommendations matched to your facial structure
🪞 AR Try-On	Virtually try styles on your webcam feed using AI-generated overlays
🤖 AI Chatbot	Built-in rule-based chatbot to guide users
📜 Analysis History	Every analysis is saved to your personal dashboard
🔐 Auth System	Secure register/login with bcrypt password hashing
🎭 Auto Gender Detection	Gemini Flash Lite auto-detects gender from the uploaded photo

---

🛠️ Tech Stack

Backend    : Python 3.10+, Flask 2.3, Gunicorn
AI/CV      : MediaPipe 0.10, OpenCV (headless), TensorFlow 2.13, DeepFace
APIs       : Google Gemini Flash Lite (gender detection), Replicate (AR style generation)
Database   : MongoDB (Atlas for cloud, localhost for dev)
Auth       : Flask-Bcrypt, Flask-Login
Frontend   : Vanilla HTML/CSS/JS (Jinja2 templates)
Deployment : Render (render.yaml included)

---

🖥️ Screenshots

Upload your photo → get your face shape → see personalised style recommendations → try them on live via AR.

---

🚀 Getting Started (Local)

Prerequisites

• Python 3.10+
• MongoDB running locally (mongod)
• A Gemini API key (get one free)
• A Replicate API token (get one free)

1. Clone the repo

git clone https://github.com/YOUR_USERNAME/hairstyle-ai.git
cd hairstyle-ai

2. Create a virtual environment

python -m venv venv

# Windows
venv\Scripts\activate

# macOS / Linux
source venv/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Set up environment variables

cp .env.example .env

Edit .env and fill in your values:

SECRET_KEY=your_random_secret_key
MONGO_URI=mongodb://localhost:27017/hair_beard_ai
GEMINI_API_KEY=your_gemini_api_key
REPLICATE_API_TOKEN=your_replicate_token
DEBUG=True

5. Run the app

python app.py

Open http://localhost:5000 in your browser.

---

☁️ Deploying to Render

This repo includes a render.yaml for one-click deployment.

1. Push this repo to GitHub
2. Go to render.com → New → Web Service
3. Connect your GitHub repo
4. Render auto-detects render.yaml — review settings and confirm
5. Add these Environment Variables in the Render dashboard:

Variable	Value
SECRET_KEY	Any random 32-char string
MONGO_URI	Your MongoDB Atlas connection URI
GEMINI_API_KEY	Your Gemini API key
REPLICATE_API_TOKEN	Your Replicate token
DEBUG	False

6. Click Deploy — your app will be live in ~3 minutes ✅

MongoDB Atlas: For cloud deployment, use a free MongoDB Atlas cluster. Make sure to whitelist 0.0.0.0/0 in Network Access.

---

📁 Project Structure

hairstyle-ai/
│
├── app.py                  # Application factory
├── wsgi.py                 # Gunicorn/WSGI entry point
├── config.py               # Configuration (reads from .env)
├── database.py             # MongoDB connection module
│
├── routes/
│   ├── auth.py             # Register, Login, Logout
│   ├── analysis.py         # Face analysis, AR snapshot, history
│   └── chatbot.py          # AI chatbot API endpoint
│
├── utils/
│   ├── face_utils.py       # MediaPipe face landmark analysis
│   ├── gender_detector.py  # Gemini-based gender auto-detection
│   ├── style_preview.py    # AR Try-On rendering (Replicate + OpenCV)
│   └── style_recommender.py# Face-shape → style mapping engine
│
├── templates/              # Jinja2 HTML templates
├── static/                 # CSS, JS, images, uploads
│
├── requirements.txt        # Python dependencies
├── Procfile                # Render/Heroku start command
├── render.yaml             # Render deployment config
└── .env.example            # Environment variable template

---

🔐 Environment Variables

Variable	Required	Description
SECRET_KEY	✅	Flask session secret (use a long random string)
MONGO_URI	✅	MongoDB connection URI
GEMINI_API_KEY	✅	Google Gemini API key for gender detection
REPLICATE_API_TOKEN	✅	Replicate token for AR image generation
UPLOAD_FOLDER	❌	Default: static/uploads
MAX_CONTENT_LENGTH	❌	Default: 16777216 (16 MB)
DEBUG	❌	Default: False (set True for local dev)

---

👤 Author

S. Thiruselvam (Roll No: 23COS263)
AI-powered personal styling assistant — built as a final-year project using Python, Flask, MediaPipe, and MongoDB.

---

📄 License

This project is for educational purposes. All rights reserved © 2024 S. Thiruselvam.