Classification of Area of Interest based on 2D map using Segmentation for path planning of Airborne Landmines Detection

There are many countries that face problems from the buried landmines in their territories. Undetected landmines cause issues to the citizens and economies of those countries. Therefore, we need to come up with cost-effective solutions to detect landmines. There are many Landmines detection techniques based on image processing. Those images are usually captured by technologies such as ground penetration radar (GPR), infrared (IR), and ultrasound (US) sensing or magnetometry imaging. Image processing techniques, including filtering, enhancement, feature extraction, and segmentation, are applied to captured images to automate the detection process. Combined with ML and AI techniques, Airborne sensing using UAV technology is sought to reduce the cost and improve on detection performance. In this paper, we present aspects of the ongoing effort to develop an automated detection system based on an integrated approach that utilizes AI and UAV technologies. In this work, we focus on developing segmentation models based on 2D maps to identify an area of interest where the UAV needs to traverse to capture the images. The automatic classification of the area of interest will then be used to plan the path that UAVs will use to traverse the area by looking at the magnetic signature of the landmines over such sites. We have created our dataset and compared the performance using different segmentation models. The results are satisfactory, and we can utilize the proposed models to achieve the objective.