International Journal of Multimedia Computing, 2025, 6(1); doi: 10.38007/IJMC.2025.060105.
GuangJun Lai, HaiMin Wang, Huiheng Suo, TingQi Zhou, Meng Qin, Qingyuan Xiao, Zuteng Chen, Jian Wu, Yuanhao Pan, Yingping Bai, Qinxi Lin
Nanchang Hangkong University, Nanchang, China
Aiming at the key technical challenges of small target detection in UAV aerial photography scenarios, this study proposes an improved scheme SPD-YOLO based on the YOLOv8n architecture. The scheme achieves performance breakthroughs through three core innovative modules: 1) adopting the SPD-Conv module instead of the traditional downsampling operation to maintain the resolution of the feature map through the spatial pyramid decomposition strategy; 2) introducing the P2 high-resolution feature layer to construct an enhanced feature pyramid, which improves the feature extraction capability of tiny targets; 3) adopting the WIoU v3 loss function to optimize the positioning accuracy through the dynamic focusing mechanism. Experiments on the VisDrone2019 test set demonstrate that the complete solution (SPD-Conv + P2 + WIoUv3) achieves an mAP@0.5 of 38.3%, surpassing the baseline YOLOv8n by 5.3 percentage points, with precision and recall reaching 49.1% and 37.2%, respectively. Ablation experiments validate the effectiveness of each module: the introduction of the P2 feature layer alone improves 2.6 percentage points, combined with WIoU v3 improves another 1.2 percentage points, and finally the introduction of the SPD-Conv module improves the overall performance by 5.3 percentage points. This scheme significantly improves the detection performance of small targets in UAV aerial photography scenarios while maintaining real-time detection speed.
small target detection; SPD-Conv; UAV aerial photography; YOLOv8n; WIoUv3
GuangJun Lai, HaiMin Wang, Huiheng Suo, TingQi Zhou, Meng Qin, Qingyuan Xiao, Zuteng Chen, Jian Wu, Yuanhao Pan, Yingping Bai, Qinxi Lin. SPD-YOLO: Enhanced Small Object Detection for Drone Imagery. International Journal of Multimedia Computing (2025), Vol. 6, Issue 1: 56-63. https://doi.org/10.38007/IJMC.2025.060105.
[1] Uchinoura S ,Miyao J ,Kurita T .An Object Detection Method Using Probability Maps for Instance Segmentation to Mask Background:Regular Papers[J].Journal of Advanced Computational Intelligence and Intelligent Informatics,2023,27(5):886-895.
[2] Tang X ,Jia C ,He Z .UAV Path Planning: A Dual-Population Cooperative Honey Badger Algorithm for Staged Fusion of Multiple Differential Evolutionary Strategies[J].Biomimetics,2025,10(3):168-168.
[3] Leilei F ,Jun Y ,Zhiyi H .SSD Object Detection Algorithm Based on Feature Fusion and Channel Attention[J].International Journal of Advanced Network, Monitoring and Controls,2022,7(3):80-89.
[4] Su J ,Song Z ,Wang X , et al.Maize Seedling and Weed Detection Using BFSL-YOLOv8[J].World Scientific Research Journal,2025,11(3):20-28.
[5] Han J ,Chen H ,Ding Y , et al.You Only Look Once–Aluminum: A Detection Model for Complex Aluminum Surface Defects Based on Improved YOLOv8[J].Symmetry,2025,17(5):724-724.
[6] Wang X ,Xiang X .DBF-YOLO: a fall detection algorithm for complex scenes[J].Signal, Image and Video Processing,2025,19(7):532-532.
[7] Firdiantika M I ,Kim S .IS-YOLO: A YOLOv7-based Detection Method for Small Ship Detection in Infrared Images With Heterogeneous Backgrounds[J].International Journal of Control, Automation and Systems,2024,22(11):3295-3302.
[8] Zhao X ,Zhang H ,Zhang W , et al.MSUD-YOLO: A Novel Multiscale Small Object Detection Model for UAV Aerial Images[J].Drones,2025,9(6):429-429