Traffic Sign Segmentation and Classification

ML Final Project CS451

Ben Vandenbosch and Henry Mound

In order for this code to run, the user must have Pillow, TensorFlow, and OpenCV2 to run. create_dataset.py reads from the Berkeley DeepDrive dataset which can be downloaded here. In order for the code to run, the bdd100k_images and bdd100k_images folders should be in the traffic-lights directory.

To run an image through yolo_opencv.py, use the following format:

python yolo_opencv.py --image IMAGE.jpg --config yolov3.cfg --weights yolov3.weights --classes yolov3.txt

Credit to:

• Arun Ponnusamy
• Daniel Kalaspuffar