This project recognize traffic lights at korea. It uses YOLOv3 and OpenCV libraries with python3.
It takes 7,000 traffic light images and 30,000 training counts for training.
We used 10 classes. 3balls traffic lights or 4balls traffic lights.
| class number | class name |
|---|---|
| 0 | 3 red |
| 1 | 3 yellow |
| 2 | 3 green |
| 3 | 3 left |
| 4 | 4 red |
| 5 | 4 green |
| 6 | 4 yellow |
| 7 | 4 red left |
| 8 | 4 left green |
| 9 | 4 red yellow |
terminal
python3 main.py
and go url with your browsers.
http://localhost/
done!
