Addressing the challenges of low detection precision and excessive parameter volume presented by the high resolution, significant scale variations, and complex backgrounds in UAV aerial imagery, this paper introduces MFP-YOLO, a lightweight detection algorithm based on YOLOv5s. Initially, a multipath inverse residual module is designed, and an attention mechanism is incorporated to manage the issues associated with significant scale variations and abundant interference from complex backgrounds. Then, parallel deconvolutional spatial pyramid pooling is employed to extract scale-specific information, enhancing multi-scale target detection. Furthermore, the Focal-EIoU loss function is utilized to augment the model’s focus on high-quality samples, consequently improving training stability and detection accuracy. Finally, a lightweight decoupled head replaces the original model’s detection head, accelerating network convergence speed and enhancing detection precision. Experimental results demonstrate that MFP-YOLO improved the mAP50 on the VisDrone 2019 validation and test sets by 12.9% and 8.0%, respectively, compared to the original YOLOv5s. At the same time, the model’s parameter volume and weight size were reduced by 79.2% and 73.7%, respectively, indicating that MFP-YOLO outperforms other mainstream algorithms in UAV aerial imagery detection tasks.

Addressing the challenge of efficiently detecting objects in ultra-high-resolution images during object detection tasks, this paper proposes a novel method called MegaDetectNet, which leverages foreground image for large-scale resolution image object detection. MegaDetectNet utilizes a foreground extraction network to generate a foreground image that highlights target regions, thus avoiding the computationally intensive process of dividing the image into multiple sub-images for detection, and significantly improving the efficiency of object detection. The foreground extraction network in MegaDetectNet is built upon the YOLOv5 model with modifications: the large object detection head and classifier are removed, and the PConv convolution is introduced to reconstruct the C3 module, thereby accelerating the convolution process and enhancing foreground extraction efficiency. Furthermore, a Res2Rep convolutional structure is developed to enlarge the receptive field and improve the accuracy of foreground extraction. Finally, a foreground image construction method is proposed, fusing and stitching foreground target regions into a unified foreground image. This approach replaces multiple divided sub-images with a single foreground image for detection, reducing overhead time. The proposed MegaDetectNet method’s effectiveness for detecting objects in ultra-high-resolution images is validated using the publicly available DOTA dataset. Experimental results demonstrate that MegaDetectNet achieves an average time reduction of 83.8% compared to the sub-image division method among various commonly used object detectors, with only a marginal 8.7% decrease in mAP (mean Average Precision). This validates the practicality and efficacy of the MegaDetectNet method for object detection in ultra-high-resolution images.