Everybody, regardless of their intentions, may access drones thanks to increased technologies and privatisation. The disruption of aviation order, terrorist assaults, invasions of personal privacy, and the transit of hazardous materials are thus most at risk. The threat placed should be de-escalated right away. One of the most efficient ways to cope with it is drone detection and localisation. The YOLO algorithm, one of several deep learning algorithms, applies a single neural network to the entire image, splits it into parts, and forecasts bounding boxes and probabilities for each zone. The projected probabilities are used to weight these bounding boxes. Because of its speed, FPS (Frames Per Second), and memory usage, the new version YOLOv5 is popular. Using a Kaggle drone dataset, a trained model is created. The YOLOv5 technique is then used, and the model weights are obtained. Graphs are used for trained weight interpretation. These weights are employed in the drone detection process. Images and videos are used in this proposed work to detect drones. The aim of the proposed work is to improve the YOLOv5 algorithm's ability to detect drones with greater precision.