Unreal Engine'de çok modlu drone tespit verisi üretimi

In recent years, the rapid advancement of drone technology has resulted in significant challenges for security and surveillance systems. Effective drone detection methods are crucial for various applications, including national security, private property protection, and safe airspace management. This thesis addresses the need for comprehensive datasets required to train robust drone detection models by leveraging simulation environments. Our approach utilizes Unreal Engine and AirSim plugin in conjunction with the AirSim GUI, a simulation environment created to generate multi-modal datasets that include both visual and audio data of drones in diverse scenarios. The primary objective of this research is to create a cost-effective and scalable method for generating high-quality datasets to improve the performance of drone detection models. We simulate a drone in a custom-built environment within Unreal Engine, capturing images and audio clips to form the dataset. The AirSim GUI allows for precise control over the simulation parameters, including the ability to spawn drones at specific coordinates, integrate various types of cameras, define flight paths, and record audio. We implemented an audio generation process that includes both real-time and offline capabilities, ensuring comprehensive coverage of possible detection scenarios. Our results demonstrate that the synthetic data generated through this method enhances the performance of drone detection models, particularly in situations where real-world data is scarce.