Smart Drone-Based Surveillance System for Real-Time Dog Detection and Alerts
C. M. Naveen Kumar, S Pradeep, Ananda Babu Jayachandra, H M Darshan, Charan Gowda H V, Kola Srinivas · 2025
The integration of drone technology has significantly advanced surveillance systems, offering unique capabilities for real-time monitoring and detection. This paper presents an efficient drone-based surveillance system tailored for dog detection and monitoring. Utilizing lightweight deep learning model namely You Only Look Once (YOLO), the system ensures accurate real-time identification of dogs in diverse environments, addressing challenges like dynamic backgrounds, variable lighting, and drone motion. Aerial imagery captured by the drone is processed in real-time to identify dogs with a precision. The system architecture comprises three key components: the drone unit, the detection and processing unit, and the monitoring system. Equipped with the cameras and sensors, the drone unit captures visual data while ensuring stable flight. The detection and processing unit employs Deep learning detection algorithm, optimized for edge computing to enable fast and efficient processing without external servers. Data augmentation and transfer learning enhance model performance across various scenarios of dogs. Upon detecting a dog, the alerting or monitoring mechanism shows in a stakeholders via a devices like smartphones and web interfaces. Additionally, the system is scalable and cost-effective, with a modular hardware and open-source software enabling adaptability to various use cases. In our approach we leverage machine learning models to classify the dog behavior and moment to shows the web interface system and smartphones.