Towards Large-Scale Benchmark Dataset for Remote Sensing Object Detection on Battlefield

Precise object detection allows military personnel to clearly understand their surroundings, leading to planning effective military strategies. Particularly, satellites and drones allow real-time surveillance over large areas, which is crucial for military operations. Collaboration among military, academic, and industrial institutions is required to promote innovation in military target detection. However, potential security concerns usually restrict access to real data from the military. To mitigate this issue, this paper proposes a novel approach to generate synthetic data for remote sensing object detection on the bat-tlefield with ARMA3, one of the renowned military tactic games. With our method, the data for model training can be easily generated without any manpower. To demonstrate the efficacy of our approach, we provide a detailed analysis of the examples from our method. As ARMA3 is well-known for its realistic military combat simulation, we believe our method can effectively contribute to military object detection in remote sensing.