Aquaculture plays a crucial role in global food security; however, inefficiencies in manual feeding monitoring pose a persistent challenge to optimizing fish production. This research proposes a novel AI-driven methodology employing YOLOv11-OBB, an advanced object detection model, to detect and analyze fish behavior for automated feeding optimization. By leveraging real-time video processing with aerial access, the proposed system accurately detects fish presence and movement patterns, enabling adaptive feeding strategies that minimize feed wastage and improve growth efficiency. Experimental results demonstrate not only high detection accuracy but also real-time performance, outperforming traditional methods in both precision and operational speed. The proposed approach contributes to the advancement of intelligent aquaculture management, offering a pathway toward sustainable and highly efficient fish farming practices.