Drone based Potholes detection using Machine Learning on various Edge AI devices in Real-Time

Pothole detection is an important aspect of road maintenance and safety. The presence of potholes on roads can cause damage to vehicles, create safety hazards for drivers, and increase maintenance costs. Timely detection of potholes can alert the driver for better navigation as well as the location of the potholes can be shared with the local governing body for taking counteractions. In this paper, we have detected potholes using two object detection models, YOLOv5, YOLOv7 & YOLOv8 and have employed the trained models on Edge AI Devices for real-time inference. We have compared the outcomes of YOLO in terms of accuracy, precision and recall. YOLOv8 is giving 78% of accuracy, YOLOv7 is giving 67% of accuracy and YOLOv5 is giving 70% of accuracy. These models are deployed on Raspberry Pi-4, Nvidia Jetson Nano and Intel UP Squared board. The performance of these Edge AI devices is analyzed in terms of fps and through experimental results it is observed that the Intel UP Squared board achieves higher fps. For real-time detection of potholes, live videos were collected through drones and were live streamed to the Edge AI devices through the RTMP server for real-time inference. Overall, the proposed system is highly efficient for the reliable detection of potholes under different road conditions.