learnopencv/Pothole-Detection-using-YOLOv4-and-Darknet

Pothole Detection using YOLOv4 and Darknet

This repository contains the code for Pothole Detection using YOLOv4 and Darknet blog post.

Here we train YOLOv4 and YOLOv4-Tiny models with 4 different experimental settings on a pothole detection dataset. We also run inference in real-time using the trained models.

• The jupyter_notebook directory contains the Jupyter Notebook which will run end-to-end with one click. You can either run it locally if you have CUDA and cuDNN installed. Or you can upload the notebook to Colab and run it in a GPU enabled environment.

• The custom_inference_script directory contains the customized darknet_video.py Python file. The Darknet code has been customized to show the FPS on the videos when running the inference. After cloning and building Darknet, replace the original darknet_video.py file with the one in the custom_inference_script directory.

Download the YOLOv4 Pothole dataset trained weights from here.

AI Courses by OpenCV

Want to become an expert in AI? AI Courses by OpenCV is a great place to start.