🍕 FoodVision AI

An end-to-end food classification application built with PyTorch, Flask, and a modern web interface.

Welcome to FoodVision AI, a project that demonstrates the entire machine learning lifecycle from model training and optimization to global deployment. This application can accurately classify 101 different food categories from any uploaded image.

✨ Features

• 101 Food Categories: Classifies a wide variety of dishes from the challenging Food-101 dataset.
• Optimized for Mobile: A lightweight frontend with client-side image resizing for a seamless experience on any device.
• High-Performance API: A Flask backend that serves a quantized PyTorch model, ensuring fast and efficient predictions.
• Robust Deployment: Configured for stable deployment on cloud platforms like Render, handling large model files effectively.

🛠️ Tech Stack

• Backend: Python, Flask, Gunicorn
• Machine Learning: PyTorch, torchvision
• Model Optimization: Quantization for a 4x size reduction
• Frontend: HTML, CSS, JavaScript
• Deployment: Render

🚀 Installation & Running Locally

Prerequisites

• Python 3.8+
• pip and git

Step 1: Clone the Repository

git clone https://github.com/Raja-89/FoodVision-AI.git
cd FoodVision-AI

Step 2: Set up the Backend

1. Create a Virtual Environment:

   python -m venv venv
   source venv/bin/activate  # On macOS/Linux
   .\venv\Scripts\activate   # On Windows

2. Install Dependencies:

   pip install -r requirements.txt

3. Get the Model: Our model is too large for a standard Git repository. Run the app.py file, and it will automatically download the necessary model file from a public URL.

4. Run the Flask App:

   python app.py

The application will be available at http://127.0.0.1:5000.

📂 Project Structure

.
├── app.py                      # Flask backend, API logic, and model handling
├── requirements.txt            # Python dependencies
├── Procfile                    # Render/Heroku deployment command
├── templates/
│   └── index.html              # Frontend web interface (HTML/CSS/JS)
├── food_classifier_model.pth   # The original PyTorch model
├── food_classifier_model_quantized.pth # The optimized, smaller model
├── .gitignore                  # Git ignore file
└── README.md                   # This file!

---