基于深度学习的遥感图像小目标检测技术研究
Research on Small Object Detection Technology in Remote Sensing Images Based on Deep Learning
DOI: 10.12677/csa.2025.157191, PDF,    科研立项经费支持
作者: 余 江, 杨晓青*：南昌职业大学校长办公室，江西 南昌
关键词: 深度学习；遥感图像小目标；测试技术；Deep Learning； Small Targets in Remote Sensing Images； Testing Technology
摘要: 本文针对遥感图像小目标检测中存在的特征提取困难、背景干扰严重以及检测精度和速度难以平衡等问题，深入探讨了多种创新方法。通过对相关算法如YOLOv11系列的改进，引入新型模块与机制，显著提升了小目标检测性能。在多个公开数据集及自建数据集上的实验表明，改进后的算法在小目标检测的准确率、召回率等指标上有显著提升，为遥感图像小目标检测的实际应用提供了有力的技术支持与参考。
Abstract: This article explores various innovative methods to address the difficulties in feature extraction, severe background interference, and difficulty in balancing detection accuracy and speed in small object detection in remote sensing images. By improving related algorithms such as the YOLOv11 series and introducing new modules and mechanisms, the performance of small object detection has been significantly improved. Experiments on multiple public and self-built datasets have shown that the improved algorithm significantly improves the accuracy and recall of small object detection, providing strong technical support and reference for the practical application of small object detection in remote sensing images.
文章引用：余江, 杨晓青. 基于深度学习的遥感图像小目标检测技术研究[J]. 计算机科学与应用, 2025, 15(7): 182-194. https://doi.org/10.12677/csa.2025.157191

参考文献

[1] Zou, Z., Chen, K., Shi, Z., Guo, Y. and Ye, J. (2023) Object Detection in 20 Years: A Survey. Proceedings of the IEEE, 111, 257-276. [Google Scholar] [CrossRef] 
[2] Lindeberg, T. (2012) Scale Invariant Feature Transform. Scholarpedia, 7, Article 10491. [Google Scholar] [CrossRef] 
[3] Tomasi, C. (2022) Histograms of Oriented Gradients. Computer Vision Sampler, 1-6.
[4] Cheng, G., Zhou, P., Yao, X., Yao, C., Zhang, Y. and Han, J. (2016) Object Detection in VHR Optical Remote Sensing Images via Learning Rotation-Invariant HOG Feature. 2016 4th International Workshop on Earth Observation and Remote Sensing Applications (EORSA), Guangzhou, 4-6 July 2016, 433-436. [Google Scholar] [CrossRef] 
[5] Cai, L., Shi, W., Hao, M., Zhang, H. and Gao, L. (2018) A Multi-Feature Fusion-Based Change Detection Method for Remote Sensing Images. Journal of the Indian Society of Remote Sensing, 46, 2015-2022. [Google Scholar] [CrossRef] 
[6] Peng, X., Zhong, R., Li, Z. and Li, Q. (2021) Optical Remote Sensing Image Change Detection Based on Attention Mechanism and Image Difference. IEEE Transactions on Geoscience and Remote Sensing, 59, 7296-7307. [Google Scholar] [CrossRef] 
[7] Cao, X., Zhang, Y., Lang, S. and Gong, Y. (2023) Swin-Transformer-Based YOLOv5 for Small-Object Detection in Remote Sensing Images. Sensors, 23, Article 3634. [Google Scholar] [CrossRef] [PubMed]
[8] Chen, W., Ouyang, S., Tong, W., Li, X., Zheng, X. and Wang, L. (2022) GCSANet: A Global Context Spatial Attention Deep Learning Network for Remote Sensing Scene Classification. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 15, 1150-1162. [Google Scholar] [CrossRef] 
[9] Ye, Y., Ren, X., Zhu, B., Tang, T., Tan, X., Gui, Y., et al. (2022) An Adaptive Attention Fusion Mechanism Convolutional Network for Object Detection in Remote Sensing Images. Remote Sensing, 14, Article 516. [Google Scholar] [CrossRef] 
[10] Tan, M., Pang, R. and Le, Q.V. (2020) EfficientDet: Scalable and Efficient Object Detection. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, 13-19 June 2020, 10778-10787. [Google Scholar] [CrossRef] 
[11] https://github.com/ultralytics/ultralytics
[12] Jegham, N., Chan, Y.K. and Marwan, A. (2024) Evaluating the Evolution of YOLO (You Only Look Once) Models: A Comprehensive Benchmark Study of YOLO11 and Its Predecessors. arXiv: 2411.00201v1.
[13] Xia, G., Bai, X., Ding, J., Zhu, Z., Belongie, S., Luo, J., et al. (2018) DOTA: A Large-Scale Dataset for Object Detection in Aerial Images. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, 18-23 June 2018, 3974-3983. [Google Scholar] [CrossRef] 
[14] Liu, Y., Shao, Z. and Hoffmann, N. (2021) Global Attention Mechanism: Retain Information to Enhance Channel-Spatial Interactions. arXiv: 2112.05561.