Monitoring Road Condition using Neural Network

Author(s): Ruchi Chaturvedi, Sheetal Avirkar


Potholes area unit one in all the foremost common road hazards. They are caused by water leaky into the cracks within the asphalt and so freezing throughout winter months. This makes them a lot of dangerous because these can cause severe harm to vehicles similarly as cause accidents. This paper uses deep learning to detect potholes on roads mistreatment pictures captured by cameras. It works by detecting edges in a picture that indicate changes in elevation that correspond with the form of a hole. This analysis paper uses YOLOv5s rule to sight potholes on the road. The exactness resulted is 82.6% and also the model detects the potholes within the confidence range of 0.50 to 0.91. The model is in a position to sight dry similarly as rainwater stuffed potholes. The future scope of this project is the readying of the mode for real time situations. we will use this model to sight potholes on roads for driverless vehicles. This project also can be wont to collect information of potholes with specific minimum size on roads. This dataset then is employed by information scientists to induce helpful insights such as observant the changes in the condition of roads in an exceedingly given span of time.