Analysis of Computer Vision Methods and Means for Explosive Ordnance Detection Mobile Systems

Abstract

The detection and removal of unexploded ordnance and landmines are vital for ensuring civil-ian safety, enabling the repurposing of affected land, and supporting post-conflict recovery ef-forts. Robotization plays a pivotal role in addressing the hazardous and labor-intensive nature of demining operations. The purpose of this paper is to review prominent computer vision tech-niques, particularly object detection, and analyze their applications in the specialized domain of explosive ordnance detection. An extensive review of the literature was conducted to explore the utilization of computer vision in the field of explosive ordnance detection. The study in-volved a comparative analysis of diverse imaging sensors employed for data capture. Special at-tention was given to reviewing sources that elaborate on different methods for detecting objects within the field of computer vision. Various approaches to object detection were thoroughly examined and compared. The research extensively examined the metrics and datasets used to evaluate different approaches for object detection. Possibilities of applying computer vision methods to detect explosive ordnance under the limitations of mobile platforms were studied. Directions of future research are formulated.