ABSTRACT
Landmine detection plays a crucial role in saving the lives of soldiers, the public, and animals. All countries have made great efforts to detect and prevent landmines from being buried. Landmines can be detected using sensor data and categorized based on several Machine Learning classification techniques. These techniques primarily use ground-penetrating radar images for landmine classification or recognition. The effectiveness of an algorithm depends on the buried object, depth, soil type, and composition used in the landmine. Recent developments in Machine Learning and the sub-fields of conventional Machine Learning provide new insights into landmine detection. Nevertheless, it is difficult to accurately identify landmines due to soil in homogeneities, surface irregularities, underground clutter, ringing effects, antenna coupling, etc. This chapter covers Machine Learning techniques like such as supervised and unsupervised learning, including Support Vector Machines, Neural Networks, K-Nearest Neighbors, and Hidden Markov models for detecting landmines. The chapter analyzes various sensor data and shows the experimental results from existing works. It also compares the classifier techniques with performance and discusses work utilizing Machine Learning algorithms. Finally, the chapter proposes deep learning architecture for the classification of landmines.
