Drone Detection and Classification Using YOLOv8 and Deep CNN

The proposed DDT-CNN model represents a deep learning architecture-based comprehensive system for drone detection and classification. YOLOv8 a cutting-edge object detection model, is used in the detection phase to precisely find drones in aerial footage. Following detection, three different convolutional neural network (CNN) architectures—Res Net 101, VGG-16, and a custom CNN model—are used to classify the drones into different groups. Deep feature extraction is possible with the Res Net 101 model, and performance and computational efficiency can be balanced with the VGG-16 model. Furthermore, a customized CNN model is designed to extract specific properties related to drone categorization. All the processing is performed on the customized dataset. The proposed model has been built because its performance rate is more accurate than other state-of-the-art models. For drone detection, the YOLO v8 model is used as its accuracy rate is high compared to other YOLO models and the detection speed is also fast. The model used for the classification of drones is deep CNN and its models, i.e., CNN, ResNet101, and VGG16, which have outstanding accuracy, recall, and precision rates compared to other state-of-the-art tools like Google Net, Alex Net, and LSTM.