Video surveillance systems provide means to detect the presence of potentially malicious drones in the surroundings of critical infrastructures. In particular, these systems collect images and feed them to a deep-learning classifier able to detect the presence of a drone in the input image. However, current classifiers are not efficient in identifying drones that disguise themselves with the image background, e.g., hiding in front of a tree. Furthermore, video-based detection systems heavily rely on the image’s brightness, where darkness imposes significant challenges in detecting drones. Both these phenomena increase the possibilities for attackers to get close to critical infrastructures without being spotted and hence be able to gather sensitive information or cause physical damages, possibly leading to safety threats. In this article, we propose RANGO, a drone detection arithmetic able to detect drones in challenging images where the target is difficult to distinguish from the background. RANGO is based on a deep learning architecture that exploits a Preconditioning Operation (PREP) that highlights the target by the difference between the target gradient and the background gradient. The idea is to highlight features that will be useful for classification. After PREP, RANGO uses multiple convolution kernels to make the final decision on the presence of the drone. We test RANGO on a drone image dataset composed of multiple already-existing datasets to which we add samples of birds and planes. We then compare RANGO with multiple currently existing approaches to show its superiority. When tested on images with disguising drones, RANGO attains an increase of 6.6% mean Average Precision (mAP) compared to YOLOv5 solution. When tested on the conventional dataset, RANGO improves the mAP by approximately 2.2%, thus confirming its effectiveness also in the general scenario.

Critical infrastructures (CIs), such as military bases and airports, are putting a lot of attention into defending against attacks delivered via drones by deploying drone detection systems. However, the CI area might be very large, with no-fly zones extending to regions where it might not be possible to deploy a power line for resourceful cameras. To this aim, the CI might deploy an Internet of Things (IoT)-based surveillance camera system to capture drone images. However, these IoT cameras are resource-constrained devices that cannot support the currently available detectors. In this article, we propose Light-YOLOv5, a lightweight image-based drone detector for resource-constrained cameras. We make targeted improvements to YOLOv5, including the replacement of the backbone network, the introduction of the transformer module, and the design of a parallel mixed efficient attention module (PEAM). We show that our modifications allow for reduced network size while achieving better classification than other state-of-the-art solutions. To prove these claims, we expanded an already available data set of blurred drone images by adding clear images of aircraft and birds. Since airplanes and birds are easily confused as drones by image classifiers, our addition proves the effectiveness of our solution. Experiments show that Light-YOLOv5 can achieve a very good tradeoff between performance (74.8% mAP) and efficiency (170 FPS). Compared to YOLOv5, Light-YOLOv5 improves mAP by 4.1%, reduces the number of network parameters by 15.7%, can perform detection at 170 frames per second (FPS), and achieves an average accuracy rate of 93.8%.