Food Image Classifier with Nutritional Info

This project is a deep learning-based food image classifier that predicts the type of food in an image and displays its nutritional information. It uses a VGG16 model with transfer learning and supports both a Tkinter GUI and a web-based Flask interface for user-friendly testing.

---

Demo

---

Features

• Classifies images into 11 food categories
• Displays nutrition details: Calories, Carbs, Fat, Protein
• Built-in GUI (Tkinter) and Web App (Flask)
• Visualizes confusion matrix
• Fine-tuning supported using transfer learning (VGG16)
• Easily extendable and interpretable

---

Tech Stack

• Python
• TensorFlow / Keras
• Scikit-learn
• Pillow (PIL)
• OpenCV
• Matplotlib & Seaborn
• Tkinter (for GUI)
• Flask (for web app)

---

Dataset Structure

food11/
├── train/
│   ├── apple_pie/
│   ├── cheesecake/
│   └── ...
└── test/
    ├── apple_pie/
    ├── cheesecake/
    └── ...

Supported classes: apple_pie, cheesecake, chicken_curry, french_fries, fried_rice, hamburger, hot_dog, ice_cream, omelette, pizza, sushi

---

Installation

1. Clone this repository:

git clone https://github.com/InsiyaFakhruddin/Food_Classifier.git
cd Food_Classifier

2. Install the required dependencies:

pip install -r requirements.txt

Or manually:

pip install tensorflow keras numpy scikit-learn opencv-python pillow matplotlib seaborn flask

---

How to Run

Web App (Flask)

python app.py

• Navigate to http://127.0.0.1:5000/
• Upload a food image
• View the prediction and nutritional info

Desktop GUI (Tkinter)

python main.py

• Choose and classify an image via a GUI window.

---

Training the Model

The model is trained using transfer learning on VGG16. You can retrain or fine-tune it by calling:

train_model(x_train, y_train, x_test, y_test)
fine_tune_model(model, datagen, x_train, y_train, x_test, y_test)

Model will be saved as:

• food_classifier_1.keras (initial)
• fine_tuned_food_classifier.keras (after tuning)

---

Project Structure

├── app.py                      # Flask web interface
├── main.py                     # Tkinter GUI & training/testing logic
├── model/                      # Pretrained Keras model(s)
├── food11/                     # Dataset
├── Food Categories/            # Sample images for display
├── templates/ & static/        # Web frontend files
├── training_history.pkl        # Saved training metrics
├── utilities.py                # Helper methods
├── requirements.txt            # Dependency list

---

Author

Insiya Fakhruddin
AI & Deep Learning Enthusiast
GitHub

---

License

This project is licensed under the MIT License – feel free to use, modify, and share.

---

Acknowledgements

• VGG16 from Keras Applications
• Dataset: Food-11 (or your custom collection)