Ground Fissure Detection System using Raspberry Pi and Drone Technology

Ground fissure results in a significant risk to animals, humans, and infrastructure. The conventional method for indirect georeferencing of remote sensing images cannot meet the timeliness requirements due to its processing speed. This study focuses on a Ground Fissure Detection System to efficiently provide data about existing ground fissures for thorough research of the experts. It also aims to provide a 2D mapping of the area and have approximation length and width of the fissures that will be useful for risk and evacuation planning. The detection accuracy and speed of the system are based on the evaluation and training of models, and testing in an actual site in Lemery, Batangas. The SSD MobileNet algorithm was used for fissure detection with detection accuracy of 59.37% at 39ms processing speed.