Abstract
Artificial Intelligence has grown dramatically in the last couple of years because of many advanced Deep learning models that use Natural Language processing, Image processing, Computer vision, etc. Drones are Technically becoming the mainstream of the workforce. Using these deep learning machine learning algorithms paired with drones, we can exponentially explore problems in the field in real-time. Droneguard is introducing a comprehensive system that combines deep learning with autonomous drone technology for the real-time identification and monitoring of power infrastructure and data mapping. We propose a custom YOLO model (you only look once), which is much faster than traditional models like R-CNN, Faster R_CNN, RPNN, etc.
This model, trained on a dataset specifically annotated for power infrastructure detection, processes aerial footage to detect key infrastructure elements. Along with the YOLO model, we have an application where the system integrates Geotagging the locations, representing them on real-time maps using GPS modules on the drones, and this ensures future inspections or maintenance operations of the specific target areas with high precision. DRONEGUARD automates both the detection and navigation processes for infrastructure inspection and reduces the time, cost, and manual effort. Choosing hardware necessary, like Raspberry Pi with Coral TPU, Nvidia Jetson nano developer boards, which can run the program in Real Time.
We develop the DRONEGUARD system by implementing deep learning with computer vision and adding features like geospatial mapping and autonomous navigation. This offers autonomous infrastructure inspection and monitoring. This project is a scalable and efficient solution that saves a lot of human resources and costs.