<![CDATA[Perimeter monitoring in agricultural fields is essential for maintaining security and ensuring continuous observation of field conditions. This study develops a real-time human–drone interaction system using hand-gesture recognition based on MediaPipe Hands and a Support Vector Machine (SVM) classifier. A custom dataset of 24,000 images across 12 gesture classes was collected and converted into 42 hand landmarks (x, y, z), normalized relative to the wrist point. The SVM model with an RBF kernel was trained using an 80:20 split and achieved a testing accuracy of 99.18%. The system operates at 109 FPS with an average latency of 9.16 ms, enabling rapid and reliable drone responses to gesture commands. Field testing in a cornfield with FPV camera visualization demonstrated that the system consistently recognized gestures in varying outdoor lighting, allowing drones to execute precise perimeter checks and maneuvers. These results highlight the significant potential of integrating gesture recognition with drone control, providing a practical, real-world solution that advances smart farming, increases agricultural efficiency, and supports technological progress toward Sustainable Development Goals. The proposed system thus offers a lightweight, responsive, and impactful tool for modern agricultural perimeter monitoring.]]>
Copyrights © 2026
