Waste Classification AI

This project is an AI-powered waste classification system designed to identify types of waste using image classification. The system includes a trained model, a Flask backend API for classification, and a React Native frontend for uploading or capturing images for classification. The model was trained using TensorFlow and deployed locally for demonstration purposes.

---

Features

• Image Classification: Classifies waste into 12 categories:

  • Battery
  • Biological
  • Brown-glass
  • Cardboard
  • Clothes
  • Green-glass
  • Metal
  • Paper
  • Plastic
  • Shoes
  • Trash
  • White-glass

• Mobile App:

  • Upload images from the gallery.
  • Capture images using the device camera.
  • Displays the top classification result with the highest confidence score.

• Backend:

  • Flask API for classification.
  • TensorFlow model integration.
  • Handles image preprocessing and prediction.

---

Tech Stack

Frontend

• React Native (Expo)
• Expo Image Picker for gallery and camera access

Backend

• Flask
• TensorFlow
• Flask-CORS

Deployment

• Local deployment for demo purposes
• Railway (optional for hosting the Flask backend)

---

Installation and Setup

Prerequisites

• Node.js and npm installed
• Python 3.9+
• Expo CLI installed globally:

  npm install -g expo-cli

• Install Python dependencies:

  pip install -r requirements.txt

---

Backend Setup(Already deployed on Railway, only neccesary if you want to train or test locally)

1. Clone the repository:

   git clone https://github.com/your-repo/waste-classification
   cd waste-classification/backend

2. Install dependencies:

   pip install -r requirements.txt

3. Run the Flask API locally:

   python app.py

4. Test the API using Postman or curl:

   curl -X POST -F [email protected] http://127.0.0.1:5000/classify

---

Frontend Setup

1. Navigate to the frontend directory:

   cd ../frontend

2. Install dependencies:

   npm install

3. Start the Expo server:

   npx expo start

4. Download the Expo Go app on your Iphone and scan the QR code using the Iphone camera to view the app.

---

Usage

1. Open the React Native app on your iPhone using Expo Go.
2. Select an image from the gallery or take a picture.
3. Press the "Classify Image" button.
4. View the classification result, showing the type of waste with the highest confidence score.

---

Model Information

• Architecture: MobileNetV2 pretrained on ImageNet with custom classification layers.
• Training Data: Kaggle Garbage Classification dataset.
• Output: Trained TensorFlow model (waste_classified_epoch_10_val_accuracy_0.94.h5).

---

Known Issues

• Misclassifications may occur due to insufficient training data for certain categories.
• Aluminum cans may be classified as plastic due to dataset limitations.
• Requires local Flask backend for the React Native app to work.

---

Improvements and Future Work

1. Dataset Expansion:
   • Include more samples for underrepresented categories (e.g., aluminum cans).
2. Model Optimization:
   • Fine-tune the model with additional epochs and data augmentation.
3. Deployment:
   • Host the backend on a cloud platform like Railway or AWS.
   • Build and deploy the React Native app using Expo EAS for a standalone iOS app.

---

Folder Structure

waste-classification/
├── backend/
│   ├── app.py
│   ├── requirements.txt
│   ├── waste_classified_epoch_10_val_accuracy_0.94.h5
├── frontend/
│   ├── App.js
│   ├── package.json
│   ├── node_modules/
├── README.md

---

License

This project is licensed under the MIT License.