Benchmarking Deep Learning and Statistical Target Detection Methods for PFM-1 Landmine Detection in UAV Hyperspectral Imagery

A new benchmark study compared deep learning against four classical statistical methods for detecting PFM-1 landmine targets in UAV hyperspectral imagery. While the Adaptive Cosine Estimator (ACE) method achieved the highest ROC-AUC score of 0.989, a proposed lightweight Spectral Neural Network outperformed all classical detectors in precision-focused metrics like Average Precision. The research highlights the need for precision-focused evaluation in sparse target scenarios and releases pixel-level ground truth data for standardized future testing.