Accurate drone detection in aerial imagery is challenging due to small object size, and environmental factors. Existing methods often struggle with robustness and adaptability across varied real-world scenarios. AI-driven drone detection system using computer vision is developed to enhance surveillance in restricted zones. A custom high-resolution dataset was created by recording drone flights using a 60 FPS high-resolution camera. Every 10th frame was extracted, resulting in 6,126 images (3840×2160), encompassing 18,603 drone instances. All images were manually annotated to ensure precision and quality. The YOLOv11n model was employed for training due to its efficiency and real-time detection capabilities. This solution addresses the limitations of traditional radar and acoustic systems, offering a more accurate and scalable approach for airspace monitoring.