| 100 | 0 | 75 |
| 下载次数 | 被引频次 | 阅读次数 |
针对遥感拍摄目标角度变化导致图像中的检测目标尺度多样且密集分布难以准确检测的问题，提出一种多元分支融合自注意力（Multi Branch Fusion Self-Attention,MFS）的遥感图像目标检测算法。该算法先设计由卷积和自注意力机制组成的多分支模块，形成特征提取网络，再建立针对小物体的第4个检测头，旨在融合不同尺度的特征。同时，利用DepGraph剪枝方法进行剪枝，降低参数规模使其轻量化。实验结果表明，所提算法在航拍图像（Dataset for Object deTection in Aerial Image,DOTA）数据集和NWPU VHR-10(Northwestern Polytechnical University Very High Resolution-10)数据集的平均准确率分别为77.7%和96.5%，优于同等参数规模的检测算法。特别是在剪枝后，参数规模仅有6.64M的情况下，所提算法对DOTA数据集检测精度可以保持在72.9%。
Abstract：To address the challenges of scale variation and dense object distribution in remote sensing imagery caused by varying imaging angles,a novel object detection algorithm is proposed based on multi-branch fusion self-attention(MFS).A multi-branch module that integrates convolutional and self-attention mechanisms is designed to build a feature extraction network,and the fourth detection head is built for small objects to facilitate multi-scale feature fusion.Meanwhile,the resulting model is pruned by the DepGraph method to achieve a lightweight architecture.Experiments on the DOTA and NWPU VHR-10 datasets demonstrate that the proposed algorithm achieves mean average precision(mAP)scores of 77.7%and 96.5%respectively,outperforming the peer detectors of similar algorithm complexity.Notably,the pruned version maintains a mAP of 72.9% on DOTA,with only 6.64 million parameters.
[1] CHENG G,HAN J W.A survey on object detection in optical remote sensing images[J].ISPRS Journal of Photogrammetry and Remote Sensing,2016,117:11-28.
[2] WANG H N,LI Y,FANG Y Q,et al.SRS-Net:Training object detectors from scratch for remote sensing images without pretraining[J].Chinese Journal of Aeronautics,2023,36(8):269-283.
[3]黄泽贤，吴凡路，傅瑶，等．基于深度学习的遥感图像舰船目标检测算法综述[J]．光学精密工程，2023,31(15):2295-2318.HUANG Z X,WU F L,FU Y,et al.Review of deep learning-based algorithms for ship target detection from remote sensing images[J].Optics and Precision Engineering,2023,31(15):2295-2318.(in Chinese)
[4] ZHANG Y,SONG C L,ZHANG D W.Small-scale aircraft detection in remote sensing images based on Faster-RCNN[J].Multimedia Tools and Applications,2022,81(13):18091-18103.
[5]张哲益，曹卫华，朱蕊，等．基于脉冲卷积神经网络稀疏表征的高分辨率遥感图像场景分类方法[J]．控制与决策，2022,37(9):2305-2313.ZHANG Z Y,CAO W H,ZHU R,et al.Sparse representation with spike convolutional neural networks for scene classification of remote sensing images of high resolution[J].Control and Decision,2022,37(9):2305-2313.(in Chinese)
[6]谢星星，程塨，姚艳清，等．动态特征融合的遥感图像目标检测[J]．计算机学报，2022,45(4):735-747.XIE X X,CHENG G,YAO Y Q,et al.Dynamic feature fusion for object detection in remote sensing images[J].Chinese Journal of Computers,2022,45(4):735-747.(in Chinese)
[7]周葳楠，吴治海，张正道，等．基于弱特征增强的轻量化小目标检测方法[J]．控制与决策，2024,39(2):381-390.ZHOU W N,WU Z H,ZHANG Z D,et al.Lightweight small target detection method based on weak feature enhancement[J].Control and Decision,2024,39(2):381-390.(in Chinese)
[8]严春满，王铖．基于特征增强的SAR图像舰船小目标检测算法[J]．控制与决策，2023,38(1):239-247.YAN C M,WANG C.A ship small target detection algorithm based on feature enhancement in SAR image[J].Control and Decision,2023,38(1):239-247.(in Chinese)
[9]田中原．遥感图像多尺度目标的轻量化检测方法[J]．测绘科学，2023,48(6):104-111.TIAN Z Y.Lightweight detection method for multiscale objects in remote sensing images[J].Science of Surveying and Mapping,2023,48(6):104-111.(in Chinese)
[10]AHMED M,EL-SHEIMY N,LEUNG H,et al.Enhancing object detection in remote sensing:A hybrid YOLOv7 and transformer approach with automatic model selection[J].Remote Sensing,2024,16(1):51-68.
[11]赵同祥，张瑞全，高树静，等．基于DN-YOLOv5遥感目标快速检测方法[J]．电子设计工程，2024,32(4):186-190.ZHAO T X,ZHANG R Q,GAO S J,et al.A fast detection method of remote sensing target based on DNYOLOv5[J].Electronic Design Engineering,2024,32(4):186-190.(in Chinese)
[12]庄文华，唐晓刚，张斌权，等．基于改进YOLOv5的遥感图像旋转框目标检测[J]．电子设计工程，2023,31(14):137-141.ZHUANG W H,TANG X G,ZHANG B Q,et al.Remote sensing image rotatable bounding box object detection based on improved YOLOv5[J].Electronic Design Engineering,2023,31(14):137-141.(in Chinese)
[13]兰旭婷，郭中华，石甜甜，等．融合SPP与FPN的光学遥感图像飞机目标检测[J]．电光与控制，2023,30(4):6-11.LAN X T,GUO Z H,SHI T T,et al.Aircraft target detection in optical remote sensing images by fusing SPP and FPN[J].Electronics Optics&Control,2023,30(4):6-11.(in Chinese)
[14]WANG C Y,BOCHKOVSKIY A,LIAO H M.YOLOv7:Trainable bag-of-freebies sets new state-ofthe-art for real-time object detectors[C]//2023IEEE/CVF Conference on Computer Vision and Pattern Recognition.Vancouver:IEEE,2023:7464-7475.
[15]LIU S,QI L,QIN H F,et al.Path aggregation network for instance segmentation[C]//2018IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:8759-8768.
[16]DING X H,ZHANG X Y, MA N N,et al.RepVGG:Making VGG-style ConvNets great again[C]//2021IEEE/CVF Conference on Computer Vision and Pattern Recognition.Nashville:IEEE,2021:13728-13737.
[17]ZHANG H,XU C,ZHANG S J.Inner-IoU:More effective intersection over union loss with auxiliary bounding box[EB/OL].[2024-10-13].https：//arxiv.org/abs/2311.02877.
[18]FANG G F,MA X Y,SONG M L,et al.DepGraph:Towards any structural pruning[C]//2023IEEE/CVF Conference on Computer Vision and Pattern Recognition.Vancouver:IEEE,2023:16091-16101.
[19]LIU Z,LI J G,SHEN Z Q,et al.Learning efficient convolutional networks through network slimming[C]//2017IEEE International Conference on Computer Vision.Venice:IEEE,2017:2755-2763.
[20]XIA G S,BAI X,DING J,et al.DOTA:A largescale dataset for object detection in aerial images[C]//2018IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:3974-3983.
[21]CHENG G,HAN J W,ZHOU P C,et al.Multi-class geospatial object detection and geographic image classification based on collection of part detectors[J].ISPRS Journal of Photogrammetry and Remote Sensing,2014,98:119-132.
[22]RAN Q,WANG Q,ZHAO B Y,et al.Lightweight oriented object detection using multiscale context and enhanced channel attention in remote sensing images[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14:5786-5795.
[23]ZHU X K,LYU S C, WANG X,et al.TPHYOLOv5:Improved YOLOv5 based on transformer prediction head for object detection on drone-captured scenarios[C]//2021IEEE/CVF International Conference on Computer Vision Workshops. Montreal:IEEE,2021:2778-2788.
[24]LIU W,ANGUELOV D,ERHAN D,et al.SSD:Single shot MultiBox detector[C]//Computer VisionECCV 2016.Cham:Springer,2016:21-37.
[25]LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detection[C]//2017IEEE International Conference on Computer Vision.Venice:IEEE,2017:2999-3007.
[26]TERVEN J,CóRDOVA-ESPARZA D M,ROMERO-GONZáLEZ J A.A comprehensive review of YOLO architectures in computer vision:From YOLOv1to YOLOv8and YOLO-NAS[J].Machine Learning and Knowledge Extraction,2023,5(4):1680-1716.
[27]HUANGFU P P,DANG L X.A multi-scale pyramid feature fusion-based object detection method for remote sensing images[J].International Journal of Remote Sensing,2023,44(24):7790-7807.
[28]ZHENG Z,ZHONG Y F,MA A L,et al.HyNet:Hyper-scale object detection network framework for multiple spatial resolution remote sensing imagery[J].ISPRS Journal of Photogrammetry and Remote Sensing,2020,166:1-14.
[29]吕奕龙，李敏，吴肇青，等．稠密连接递归特征金字塔的遥感目标检测算法[J]．遥感学报，2024,28(6):1602-1614.LYU Y L,LI M,WU Z Q,et al.Object detection in remote sensing images using densely connected recursive feature Pyramids[J].National Remote Sensing Bulletin,2024,28(6):1602-1614.(in Chinese)
[30]GE Z,LIU S T,WANG F,et al.YOLOX:Exceeding YOLO series in 2021[EB/OL].[2024-10-13].https：//arxiv.org/abs/2107.08430.
[31]LIU J P,ZHENG K Y,LIU X Y,et al.SDSDet:A real-time objectdetector for small,dense,multi-scale remote sensing objects[J].Image and Vision Computing,2024,142:104898-104913.
[32]刘富宽，罗素云，何佳，等．FVIT-YOLO v8：基于多尺度融合注意机制的改进YOLO v8小目标检测[J]．红外技术，2024,46(8):912-922.LIU F K,LUO S Y,HE J,et al.FVIT-YOLOv8:Improved YOLOv8 small object detection based on multi-scale fusion attention mechanism[J].Infrared Technology,2024,46(8):912-922.(in Chinese)
[33] WANG P J,SUN X,DIAO W H,et al.FMSSD:Feature-merged single-shot detection for multiscale objects in large-scale remote sensing imagery[J].IEEE Transactions on Geoscience and Remote Sensing,2020,58(5):3377-3390.
[34]MIN L T,FAN Z M,LV Q Y,et al.YOLO-DCTI:Small object detection in remote sensing base on contextual transformer enhancement[J].Remote Sensing,2023,15(16):3970-3988.
基本信息:
DOI：10.13682/j.issn.2095-6533.2025.06.011
中图分类号:TP751
引用信息:
[1]亢红波,温家正,杨春杰,等.多元分支融合自注意力的遥感图像目标检测算法[J].西安邮电大学学报,2025,30(06):94-103.DOI:10.13682/j.issn.2095-6533.2025.06.011.
基金信息:
陕西省科技厅重点计划项目(2018ZDXM-GY-039)