A powerful AI-based Streamlit web app that allows users to upload food images, receive calorie estimations based on image classification, and chat with Google Gemini about any food-related questions!
- 📸 Food Image Recognition – Upload an image, and get the predicted food name using a PyTorch model.
- 🔥 Calorie Estimation – Matches prediction with a dataset of 100+ foods to provide estimated calories.
- 💬 Gemini AI Chat – Ask questions about the food, ingredients, recipes, health info, and more!
- 🖥️ Streamlit Interface – Clean and interactive web UI, easy to use even without technical background.
| Tool | Purpose |
|---|---|
| PyTorch | Food image classification with ResNet18 |
| Streamlit | Interactive web interface |
| Google Gemini API | Food-related conversational AI |
| Pandas | Handling and searching food-calorie data |
| PIL | Image preprocessing |
food-ai-app/ ├── app.py # Main Streamlit application script ├── calories.csv # CSV file containing food names and calorie values ├── classes.txt # Text file containing class names for the model ├── resnet152_food21_best.pt # Trained PyTorch model (ResNet-based) ├── requirements.txt # Python dependencies ├── README.md # Project documentation (this file) └── images/ # (Optional) Folder for storing test/sample images
pip install -r requirements.txt
Get your Gemini API key from Google AI Studio, then paste it in your app.py: genai.configure(api_key="YOUR_API_KEY")
Image Upload: User uploads an image.
Image Preprocessing: The image is transformed using PyTorch's transforms.
Prediction: The ResNet18 model predicts the food label.
Calorie Match: The label is searched in calories.csv for the estimated calorie.
Chat: User can ask Gemini AI anything about the food.
Developed by Mainak Roy
This project is licensed under the MIT License. Feel free to use, modify, and share it!