Abstract: Birds are an integral part of our ecosystem and play a key role in maintaining nature. The use of drones poses a serious threat to birds. Drones can disturb bird populations and cause damage and possibly death. Drones have the potential to cause serious damage to the ecosystem and bird habitats. The development of a system capable of accurately identifying and categorizing drones and birds is essential to reducing this risk. The YOLOv4 and YOLOv5 models are going to be utilized in the construction of the model that we intend to develop for this project, which is going to be able to identify drones as well as birds. This is because the use of drones provides a considerable threat to bird populations.

Once the models have been trained, we will first determine how well they did on a test set, and then we will determine how well they did on a validation set. It is of the highest significance to develop a system that is capable of precise detection and classification of both drones and birds in light of the growing frequency of the use of drones, which constitutes a substantial threat to bird populations. The performance of the YOLOv4 and YOLOv5 models will be assessed for the purpose of detecting birds and drones, and the model that has the best overall performance will be chosen as the winner of the competition. The accuracy of this model is 94% it is estimated according to whether an object in question is a bird or a drone.

Keywords: YOLOV4, YOLOV5, CNN.

| DOI: 10.17148/IARJSET.2023.107118