Recycling
Recycling is now cheaper, offline, easier, and sustainable.
Core trick
Runs an image classifier on-device via ExecuTorch on older Android phones, mapping garbage photos to bin categories offline.
Description
Recycling is now cheaper, offline, easier, and sustainable. We use Android devices from previous years to run a sorting helper for events, closed spaces, public spaces, and any other location without requiring internet connectivity or expensive hardware.
README excerpt
# Smart Recycling Bin Assistant
An AI-powered, IoT-enabled recycling assistant that uses ExecuTorch for on-device inference to classify garbage and guide users to the correct bin.
## Project Overview
The Smart Recycling Bin Assistant uses computer vision and machine learning to:
- Classify garbage into multiple categories (paper, cardboard, glass, plastic, metal, trash)
- Provide real-time guidance on which bin to use
- Work offline with on-device inference
- Adapt to regional recycling rules via configuration
- Monitor bin fill levels and send alerts
## Project Structure
```
HackOct20/
├── README.md # This file
├── requirements.txt # Python dependencies
├── train_model.py # Local training script
├── convert_to_executorch.py # ExecuTorch conversion script
├── device_app.py # Raspberry Pi application
├── config/
│ ├── rules.json # Regional recycling rules
│ └── config.json # Device configuration
├── models/
│ ├── garbage_model.pth # Trained PyTorch model
│ ├── garbage_model.pte # ExecuTorch model
│ └── labels.txt # Class labels
├── data/
│ └── README.md # Data download instructions
└── utils/
└── helpers.py # Utility functions
```
## Quick Start
### 1. Setup Environment
```bash
# Create virtual environment
python3 -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
```
### 2. Downl