♻️ Eco Vision – Smart Waste Classifier with Azure & Streamlit

Eco Vision is a smart, AI-powered garbage classification system that helps users identify types of waste in real-time using images or camera input. Built with Azure Custom Vision, MongoDB, and Streamlit, this app not only predicts waste categories but also provides eco-friendly disposal tips, gathers user feedback, and maintains logs for continuous learning.

---

🚀 Features

• ✅ Classifies garbage into 6 categories: Plastic, Paper, Glass, Metal, Cardboard, and Trash
• 📸 Supports image upload, camera capture, and image URL input
• 🔍 Uses Azure Custom Vision API for accurate predictions
• 💬 Displays interactive disposal tips based on the prediction
• 🗳️ Collects user feedback ("Correct"/"Incorrect") to improve the model
• 📊 Logs predictions, feedback, and usage data into MongoDB
• 📚 Includes a Home Page explaining the importance of waste classification
• 🧠 Designed to be expandable — chatbot, dynamic tips, and more coming soon!

---

📦 Folder Structure

eco-vision/
├── app.py                     # Main Streamlit app
├── utils/
│   ├── azure_api.py           # Handles Azure Custom Vision prediction requests
│   └── database.py            # MongoDB logging and feedback utilities
├── requirements.txt           # Required Python packages
└── README.md                  # Project documentation

---

🧠 How It Works

1. User uploads an image, captures it from the camera, or pastes an image URL.
2. Image is sent to the Azure Custom Vision API.
3. The model returns predictions (e.g., "Plastic: 94%").
4. The app displays:
   • The prediction result
   • Friendly disposal tips
   • A feedback form
5. All activity is logged in MongoDB Atlas.

---

⚙️ Setup Instructions

1. Clone the repository

git clone https://github.com/yourusername/eco-vision.git
cd eco-vision

2. Install dependencies

pip install -r requirements.txt

3. Set your environment variables

Create a .env file in the root directory with the following content (replace placeholders with actual values):

PREDICTION_KEY=your_azure_prediction_key
ENDPOINT=your_azure_endpoint_url
PROJECT_ID=your_project_id
ITERATION_NAME=Eco_Vision
MONGODB_URI=mongodb+srv://<username>:<password>@eco-vision.mongodb.net/?retryWrites=true&w=majority&appName=eco-vision

4. Run the app

streamlit run app.py

---

🔍 Example Prediction Flow

Input: 🖼️ Image of a plastic bottle
Output:

• Predicted: Plastic (94%)
• Disposal Tip: Rinse and place in a plastic recycling bin ♻️
• Feedback Box: ✅ Correct | ❌ Incorrect
• Logs: Stored in MongoDB with timestamp, result, and feedback

---

📊 MongoDB Document Structure

Prediction Log

{
  "filename": "plastic_bottle.jpg",
  "result": [
    { "tagName": "Plastic", "probability": 0.94 }
  ],
  "timestamp": "2025-05-02T10:00:00Z"
}

Feedback Log

{
  "filename": "plastic_bottle.jpg",
  "feedback": "Incorrect",
  "timestamp": "2025-05-02T10:05:00Z"
}

---

💡 Future Enhancements

• 🔄 Retrain model with user-submitted images marked as incorrect
• 💬 Add a chatbot for FAQs & eco-tips
• 📊 Dynamic admin dashboard using Streamlit and MongoDB
• 🌍 Multilingual support for wider eco-awareness

---

🧠 Why Garbage Classification?

• Promotes recycling and reuse 🌱
• Reduces landfill impact 🗑️
• Ensures safe disposal of hazardous materials ⚠️
• Supports a cleaner ecosystem and sustainable living 🌎

---

🤝 Contributing

Pull requests are welcome! For major changes, please open an issue first.
Ideas, issues, and ⭐ stars are appreciated!

---

---

🧪 Note from the Developer (aka Me 🤓): I'm a student, and this project is running on a free-tier Azure account, so if you ever face prediction limits or occasional connection issues, don't rage-quit or cuss the app! It’s not the AI’s fault — blame the free credits 😅 I'm working on improving and scaling it as time and budget allow. Thanks for your patience and support! 💙♻️ You can check out at 🔗https://eco-vision.streamlit.app/

---

📜 License

MIT License © 2025 Sai Kumar Garlapati