RGB–Thermal UAV Dataset for Landmine Risk Mapping

This repository provides a synthetic RGB–thermal UAV dataset generated using a controlled ROS/Gazebo simulation environment for landmine risk mapping research. It contains representative synchronized RGB and thermal image pairs, simulated post-conflict terrain configurations, and annotation specifications used for evaluating landmine-related surface and near-surface risk indicators. Terrain scenarios include dry soil, sandy terrain, grassland, forest floor, rocky patches, cratered ground, and urban rubble. Ground-truth landmine risk labels were procedurally generated within the simulator to ensure consistency and reproducibility. No real-world landmine field data were used in this dataset. The dataset is intended solely for academic and non-commercial research in UAV perception, multimodal sensing, and humanitarian demining support systems.