Vision Based Machine Learning Model (YOLO) for Pothole Detection with Drones

With advancing technology and using of automatic cars, the need for detecting road problems increases. Real time image processing and machine learning based pothole detection of roads is presented suitable for road monitoring drones. The camera taken videos are converted to frames, and YOLO V۸ model, with manually labeled pictures, is used for training. Simulation results shows that the model can successfully identify potholes after the successful training of the system. Coordinates of the detected potholes are sent via SMS, or to a predefined server according to the user choice.

کلیدواژه ها:

YOLO ، Deep Learning ، Image Processing ، Pothole Detection