JoWUA

Abstract

Aerial surveillance is considered a developing sphere. One of the reasons is because the YOLO-8 model can improve both detection accuracy and speed to a great extent. The research used the YOLO-8(You Only Look Once) model to detect vehicles and humans through drone imagery in real time an application noted for its speed and accuracy under varying environmental conditions to know how effective it would be. Results shows YOLO-8 model performs better than existing methods with high level performance feature both robust accuracy and speed, an approach aimed to enhance detection accuracy and processing speed while addressing this challenge, developing a method that combines the YOLO-8 model with drone imagery for real-time detection of vehicles and people. The applied model in this greatly improve detection precision and speed, it suitable for dynamic surveillance environments. Object recognition is effective high altitude drone-based person detection. The YOLOv8 model was able to detect objects in a drone view that acted as a person detector, achieving an application for most cases of person object in the image with 0.91% accuracy. The images were captured by drones far from objects since they had a 900 × 900-pixel resolution and runtime speed of 238.41 ms, by information about these objects was very vital. This led to innovations in drone-based detection systems powered by deep learning models, YOLOv8 which is very effective in vehicle detection using drone image data.