Unmanned aerial vehicle (UAV) object detection technology is widely used in security surveillance applications, allowing for real-time collection and analysis of image data from camera equipment carried by a UAV to determine the category and location of all targets in the collected images. However, small-scale targets can be difficult to detect and can compromise the effectiveness of security surveillance. In this work, we propose a novel dual-backbone network detection method (DB-YOLOv5) that uses multiple composite backbone networks to enhance the extraction capability of small-scale targets’ features and improve the accuracy of the object detection model. We introduce a bi-directional feature pyramid network for multi-scale feature learning and a spatial pyramidal attention mechanism to enhance the network’s ability to detect small-scale targets during the object detection process. Experimental results on the challenging UAV aerial photography dataset VisDrone-DET demonstrate the effectiveness of our proposed method, with a 3% improvement over the benchmark model. Our approach can enhance security surveillance in UAV object detection, providing a valuable tool for monitoring and protecting critical infrastructure.

In recent years, unmanned aerial vehicle (UAV) image target tracking technology, which obtains motion parameters of moving targets and achieves a behavioral understanding of moving targets by identifying, detecting and tracking moving targets in UAV images, has been widely used in urban safety fields such as accident rescue, traffic monitoring and personnel detection. Due to the problems of complex backgrounds, small scale and a high density of targets, as well as mutual occlusion among targets in UAV images, this leads to inaccurate results of single object tracking (SOT). To solve the problem of tracking target loss caused by inaccurate tracking results, this paper proposes a strong interference motion target tracking method based on the target consistency algorithm for SOT based on an interframe fusion and trajectory confidence mechanism, fusing previous frames for the tracking trajectory correction of current frames, learning again from previous frames to update the model and adjusting the tracking trajectory according to the tracking duration. The experimental results can show that the accuracy of the proposed method in this paper is improved by 6.3% and the accuracy is improved by 2.6% compared with the benchmark method, which is more suitable for applications in the case of background clutter, camera motion and viewpoint change.