SmartBin: AI-Powered Waste Classification System

SmartBin is an AI-powered waste classification system that leverages deep learning models to identify and categorize different types of waste. This project provides a RESTful API built with FastAPI and supports multiple state-of-the-art deep learning models for image classification.

Model Training Source

The deep learning models used in this project are trained using the Garbage Classification repository. This repository contains a comprehensive dataset processing pipeline, model selection, training scripts, and evaluation tools to generate high-accuracy waste classification models.

🚀 Supported Deep Learning Models

SmartBin supports the following deep learning models for waste classification:

• DenseNet - densenet201
• EfficientNet - efficientnet_b7
• GoogLeNet - GoogLeNet
• LeNet - LeNet
• MobileNetV2 - mobilenet_v2
• MobileNetV3 - mobilenet_v3_large
• ResNet - resnet50
• AlexNet - AlexNet
• ShuffleNetV2 - shufflenet_v2_x2_0
• VGG - vgg
• RegNet - regnet_x_32gf

Each model processes an input image and predicts the most probable waste category. The models are optimized for both GPU and CPU execution, automatically detecting the available hardware.

---

🛠️ Deployment Guide

1. Setting Up Model Weights

Since the model weights are too large to be stored directly in the repository, we have provided a Google Drive link to download them. Follow these steps to properly set up the weights for the backend:

1. Download Model Weights

• Access the model weights from the following Google Drive link:
  Download Model Weights
• Download all files in the folder.

2. Create the weights Directory

• Navigate to the backend directory in your local project:

  cd path/to/SmartBin/backend

• Create a new folder named weights inside the backend directory:

  mkdir weights

3. Move the Weights to the weights Folder

• Place all downloaded .pth or .pt model weight files into the weights/ directory.
• Ensure that the folder structure looks like this:

  SmartBin/
  ├── backend/
  │   ├── weights/
  │   │   ├── AlexNet.pth
  │   │   ├── DenseNet.pth
  │   │   ├── ...
  

2. Now Setup for the backend and frontend (Choose one):

Option 1: Deploy with Docker

1. Install Dependencies

   • Ensure Docker and Docker Compose are installed on your system.
   • If using a GPU, install the NVIDIA Container Toolkit:

     sudo apt-get install -y nvidia-container-toolkit

2. Clone the Repository

   git clone https://github.com/AlanY1an/Smartbin
   cd smartbin

3. Build and Run Containers

   docker compose up --build -d

   This will:

   • Build the Docker images
   • Start the FastAPI backend
   • Start the frontend
   • Enable GPU acceleration if supported

4. Verify Deployment

   • Backend API: http://127.0.0.1:8000/docs
   • Frontend UI: http://127.0.0.1:3000

5. Check GPU Availability (Optional)

   docker exec -it <backend-container-id> nvidia-smi

6. Stop and Remove Containers

   docker compose down

---

Option 2: Run Locally (Without Docker)

1. Install Prerequisites

   • Python (>= 3.8)
   • Node.js (>= 18)
   • Git

2. Clone the Repository

   git clone https://github.com/your-repo/smartbin.git
   cd smartbin

3. Run the Backend

   cd backend
   pip install -r requirements.txt
   uvicorn app:app --host 0.0.0.0 --port 8000 --reload

   • The API will be available at: http://127.0.0.1:8000/docs

4. Run the Frontend

   cd frontend
   npm install
   npm run dev

   • The frontend UI will be available at: http://127.0.0.1:3000

---

📜 API Documentation

The project includes an interactive API documentation via Swagger UI:

• Open http://127.0.0.1:8000/docs in your browser.

---

📌 Future Enhancements

• Support for real-time image classification via a mobile app
• Expansion of the dataset for improved model accuracy
• Integration with AWS for scalable deployment

---

📄 License

This project is licensed under the MIT License. See LICENSE for details.