基于CDCA-YOLOv8的无人机图像小目标识别

doi:10.16257/j.cnki.1681-1070.2025.0013

摘要/Abstract

摘要： 为解决无人机航拍图像中小目标实例多、遮挡严重的问题，提出了一种新的小目标检测算法CDCA-YOLOv8。算法在骨干网络中引入了中心注意力机制，在降低计算复杂度的同时提升特征提取能力；结合可变形卷积网络的优势，改进了卷积模块，并设计了基于可变形卷积技术的C2f模块，增强多尺度特征提取。同时设计了基于自适应结构特征融合的检测头，以提高小目标检测的精度。实验表明，与YOLOv8n相比，CDCA-YOLOv8在VisDrone2019数据集上平均精度均值mAP_0.5提高了4.4个百分点，mAP_0.5∶0.95提高了3.1个百分点，展示了更优的小目标检测效果。

关键词: YOLOv8, 无人机图像, 小目标识别, 特征提取

Abstract: A new small object detection algorithm CDCA-YOLOv8 is proposed to solve the problem of multiple small object instances and severe occlusion in drone aerial images. The algorithm introduces a central attention mechanism in the backbone network, which improves feature extraction capability while reducing computational complexity. Combining the advantages of deformable convolutional networks, the convolution module is improved and a C2f module based on deformable convolution technology is designed to enhance multi-scale feature extraction. A detection head based on adaptive structural feature fusion is designed to improve the accuracy of small target detection. The experiments show that compared with YOLOv8n, CDCA-YOLOv8 improves the mean average accuracy mAP_0.5 by 4.4 percentage points on the VisDrone2019 dataset, and mAP_0.5_∶_0.95 improves by 3.1 percentage points, which demonstrates better small object detection performance.

Key words: YOLOv8, drone image, small object detection, feature extraction

中图分类号:

TN305.94

吴诗娇，林伟. 基于CDCA-YOLOv8的无人机图像小目标识别[J]. 电子与封装, 2025, 25(1): 010501 .

WU Shijiao, LIN Wei. Small Object Detection in Drone Images Based on CDCA-YOLOv8[J]. Electronics & Packaging, 2025, 25(1): 010501 .

参考文献

[1] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, Ohio, USA, 2014.
[2] HE K M, ZHANG X Y, REN S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1904-1916.
[3] GIRSHICK R. Fast R-CNN[C]// 2015 IEEE International Conference on Computer Vision (ICCV), Santiago, Chile, 2015.
[4] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 2016.
[5] REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, USA, 2017.
[6] REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. [2025-01-06]. https://arxiv.org/abs/1804.02767v1.
[7] BOCHKOVSKIY A, WANG C Y, LIAO H M. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. [2025-01-06]. https://ARXIV.org/abs/2004.10934v1.
[8] JOCHER G, LIU C Y, HOGAN A, et al. Ultralytics/yolov5: initial release[EB/OL]. [2025-01-06]. https://github.com/ultralytics/yolov5.
[9] LI C Y, LI L, JIANG H L, et al. YOLOv6: a single-stage object detection framework for industrial application[EB/OL]. [2025-01-06]. https://github.com/meituan/YOLOv6.
[10] WANG C Y, BOCHKOVSKIY A, LIAO H M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Vancouver, BC, Canada, 2023.
[11] DING X H, ZHANG X Y, MA N N, et al. Repvgg: Making vgg-style convnets great again[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2021), Nashville, Tennessee, USA, 2021.
[12] HUANGFU Z M, LI S Q. Lightweight you only look once v8: an upgraded you only look once v8 algorithm for small object identification in unmanned aerial vehicle images[J]. Applied Sciences, 2023, 13(22): 12369.
[13] 陈卫彪, 贾小军, 朱响斌, 等. 基于DSM-YOLO v5的无人机航拍图像目标检测[J]. 计算机工程与应用, 2023, 59(18): 226-233.
[14] LI Y T, FAN Q S, HUANG H S, et al. A modified YOLOv8 detection network for UAV aerial image recognition[J]. Drones, 2023, 7(5): 304.
[15] CUBUK E D, ZOPH B, MANE D, et al. AutoAugment: learning augmentation policies from data[EB/OL]. [2025-01-06]. https://arxiv.org/abs/1805.09501v3.
[16] ZHU C C, HE Y H, SAVVIDES M. Feature selective anchor-free module for single-shot object detection[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA, 2019.
[17] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]// Computer Vision – ECCV 2018, Munich, Germany, 2018.
[18] DAI J F, QI H Z, XIONG Y W, et al. Deformable convolutional networks[C]// 2017 IEEE International Conference on Computer Vision (ICCV), Venice 2017, Venice, Italy, 2017.
[19] ZHU X Z, HU H, LIN S, et al. Deformable convnets v2: More deformable, better results[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, California, USA, 2019.
[20] LIU S T, HUANG D, WANG Y H. Learning spatial fusion for single-shot object detection[EB/OL]. [2025-01-06]. https://arxiv.org/abs/1911.09516v2.

中国半导体行业协会封装分会会刊

中国电子学会电子制造与封装技术分会会刊