This is an automated script to ease the workload of creating potholes for the purpose of inserting them into a Carla map

Included in the repo is also a meshlab script to automate the meshing process of the pointcloud of the pothole for insertion in the Carla map.

The code uses the following packages and Python3 is required

• numpy
• open3d
• sklearn

The pothole.py script is responsible for reading the pointcloud of the pothole and adding a flat plane around it to simulate a road patch. For this to happen, it needs to be supplied with the folder of the .ply files from this repo as its input [https://github.com/ruirangerfan/rethinking_road_reconstruction_pothole_detection/tree/main/dataset/model1].

The script afterwards outputs the result in the completed_potholes/ and also populates the intermediate_files/ folder with the last processed pointcloud (use this for debug purposes).

The eval.py script in turn, reads the groundtruth pointcloud (the version with the road patch generated from pothole.py) and the pointcloud generated from this repo [https://github.com/Stagakis/saliency-from-pointcloud] and outputs a confusion matrix for the point classification (pothole and road classes).