《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (12): 3723-3732.DOI: 10.11772/j.issn.1001-9081.2021101802
所属专题: 人工智能
收稿日期:
2021-10-22
修回日期:
2022-01-10
接受日期:
2022-01-14
发布日期:
2022-01-19
出版日期:
2022-12-10
通讯作者:
黄朝兵
作者简介:
冯号(1996—),男,重庆人,硕士研究生,主要研究方向:信息处理、图像处理与识别基金资助:
Hao FENG1, Chaobing HUANG1(), Yuanqiao WEN2
Received:
2021-10-22
Revised:
2022-01-10
Accepted:
2022-01-14
Online:
2022-01-19
Published:
2022-12-10
Contact:
Chaobing HUANG
About author:
FENG Hao, born in 1996, M. S. candidate. His research interests include information processing, image processing and recognition.Supported by:
摘要:
YOLOv3算法被广泛地应用于目标检测任务。虽然在YOLOv3基础上改进的一些算法取得了一定的成果,但是仍存在表征能力不足且检测精度不高的问题,尤其对小目标的检测还不能满足需求。针对上述问题,提出了一种改进YOLOv3的遥感图像小目标检测算法。首先,使用K均值聚类变换(K-means-T)算法优化锚框的大小,从而提升先验框和真实框之间的匹配度;其次,优化置信度损失函数,以解决难易样本分布不均衡的问题;最后,引入注意力机制来提高算法对细节信息的感知能力。在RSOD数据集上进行实验的结果显示,与原始的YOLOv3算法、YOLOv4算法相比,所提算法在小目标“飞机(aircraft)”类上的平均精确率(AP)分别提高了7.3个百分点和5.9个百分点。这表明所提算法能够有效检测遥感图像小目标,具有更高的准确率。
中图分类号:
冯号, 黄朝兵, 文元桥. 基于改进YOLOv3的遥感图像小目标检测[J]. 计算机应用, 2022, 42(12): 3723-3732.
Hao FENG, Chaobing HUANG, Yuanqiao WEN. Remote sensing image small target detection based on improved YOLOv3[J]. Journal of Computer Applications, 2022, 42(12): 3723-3732.
算法 | mAP@0.5 | F1 | AP(aircraft) | F1(aircraft) |
---|---|---|---|---|
YOLOv3 | 0.833 | 0.825 | 0.850 | 0.850 |
YOLOv4[ | 0.862 | 0.835 | 0.864 | 0.840 |
EfficientDet[ | 0.881 | 0.858 | 0.603 | 0.620 |
文献[ | 0.903 | 0.865 | — | — |
YOLOv3[ | 0.827 | 0.835 | 0.833 | 0.830 |
YOLOv3-AKT | 0.903 | 0.870 | 0.923 | 0.900 |
表1 不同算法的检测精度比较
Tab.1 Comparison of detection accuracy among different algorithms
算法 | mAP@0.5 | F1 | AP(aircraft) | F1(aircraft) |
---|---|---|---|---|
YOLOv3 | 0.833 | 0.825 | 0.850 | 0.850 |
YOLOv4[ | 0.862 | 0.835 | 0.864 | 0.840 |
EfficientDet[ | 0.881 | 0.858 | 0.603 | 0.620 |
文献[ | 0.903 | 0.865 | — | — |
YOLOv3[ | 0.827 | 0.835 | 0.833 | 0.830 |
YOLOv3-AKT | 0.903 | 0.870 | 0.923 | 0.900 |
特征图 | 52×52 | 26×26 | 13×13 |
---|---|---|---|
Anchor[ | (8,8),(11,12),(15,14) | (18,19),(23,24),(30,32) | (40,44),(51,58),(145,178) |
Anchor-T(c=1) | (8,8),(14,15),(22,20) | (28,30),(38,40),(53,56) | (73,81),(96,109),(290,356) |
Anchor-T(c=3) | (4,4),(8,9),(15,14) | (21,22),(31,33),(46,49) | (67,74),(91,103),(290,356) |
Anchor-T(c=5) | (4,4),(7,8),(12,12) | (16,17),(23,24),(38,40) | (60,66),(84,95),(290,356) |
Anchor-T(c=7) | (4,4),(7,8),(11,11) | (15,16),(20,21),(28,30) | (40,44),(66,75),(290,356) |
Anchor-T (c=9) | (4,4),(7,7),(11,10) | (14,15),(19,20),(26,28) | (36,40),(48,54),(290,356) |
Anchor[ | (4,4),(10,11),(18,17) | (24,26),(35,36),(49,53) | (70,77),(93,106),(290,356) |
表2 不同中心框下的锚框
Tab. 2 Anchor boxes under different center anchor boxes
特征图 | 52×52 | 26×26 | 13×13 |
---|---|---|---|
Anchor[ | (8,8),(11,12),(15,14) | (18,19),(23,24),(30,32) | (40,44),(51,58),(145,178) |
Anchor-T(c=1) | (8,8),(14,15),(22,20) | (28,30),(38,40),(53,56) | (73,81),(96,109),(290,356) |
Anchor-T(c=3) | (4,4),(8,9),(15,14) | (21,22),(31,33),(46,49) | (67,74),(91,103),(290,356) |
Anchor-T(c=5) | (4,4),(7,8),(12,12) | (16,17),(23,24),(38,40) | (60,66),(84,95),(290,356) |
Anchor-T(c=7) | (4,4),(7,8),(11,11) | (15,16),(20,21),(28,30) | (40,44),(66,75),(290,356) |
Anchor-T (c=9) | (4,4),(7,7),(11,10) | (14,15),(19,20),(26,28) | (36,40),(48,54),(290,356) |
Anchor[ | (4,4),(10,11),(18,17) | (24,26),(35,36),(49,53) | (70,77),(93,106),(290,356) |
算法 | mAP@0.5 | F1 | AP(aircraft) |
---|---|---|---|
原始YOLOv3 | 0.840 | 0.850 | 0.852 |
YOLOv3[ | 0.827 | 0.835 | 0.833 |
YOLOv3[ | 0.853 | 0.855 | 0.870 |
YOLOv3-T | 0.868 | 0.860 | 0.884 |
表3 不同锚框优化算法的精度比较
Tab. 3 Accuracy comparison of different anchor box optimization algorithms
算法 | mAP@0.5 | F1 | AP(aircraft) |
---|---|---|---|
原始YOLOv3 | 0.840 | 0.850 | 0.852 |
YOLOv3[ | 0.827 | 0.835 | 0.833 |
YOLOv3[ | 0.853 | 0.855 | 0.870 |
YOLOv3-T | 0.868 | 0.860 | 0.884 |
CA模块位置 | mAP@0.5 | AP(aircraft) | F1(aircraft) |
---|---|---|---|
YOLOv3 | 0.840 | 0.850 | 0.850 |
检测头1 | 0.849 | 0.882 | 0.860 |
检测头2 | 0.846 | 0.884 | 0.860 |
检测头3 | 0.869 | 0.901 | 0.880 |
表4 不同位置插入CA的精度比较
Tab. 4 Accuracy comparison of CA inserting in different positions
CA模块位置 | mAP@0.5 | AP(aircraft) | F1(aircraft) |
---|---|---|---|
YOLOv3 | 0.840 | 0.850 | 0.850 |
检测头1 | 0.849 | 0.882 | 0.860 |
检测头2 | 0.846 | 0.884 | 0.860 |
检测头3 | 0.869 | 0.901 | 0.880 |
1 | LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 3431-3440. 10.1109/cvpr.2015.7298965 |
2 | 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. Piscataway: IEEE, 2014: 580-587. 10.1109/cvpr.2014.81 |
3 | GIRSHICK R. Fast R-CNN[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 1440-1448. 10.1109/iccv.2015.169 |
4 | REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 779-788. 10.1109/cvpr.2016.91 |
5 | REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2017:6517-6525. 10.1109/cvpr.2017.690 |
6 | REDMON J, FARHADI A. YOLOv3: an incremental improvement[EB/OL]. (2018-04-08) [2021-09-10].. 10.1109/cvpr.2017.690 |
7 | BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. (2020-04-23) [2021-09-14].. |
8 | LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020, 42(2): 318-327. 10.1109/tpami.2018.2858826 |
9 | WANG F L, SU J Y. Based on the improved YOLOV3 small target detection algorithm[C]// Proceedings of the IEEE 4th Advanced Information Management, Communicates, Electronic and Automation Control Conference. Piscataway: IEEE, 2021: 2155-2159. 10.1109/imcec51613.2021.9482076 |
10 | LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 936-944. 10.1109/cvpr.2017.106 |
11 | KISANTAL M, WOJNA Z, MURAWSKI J, et al. Augmentation for small object detection[EB/OL]. (2019-02-19) [2021-08-15].. 10.5121/csit.2019.91713 |
12 | LIU S T, HUANG D, WANG Y H. Receptive field block net for accurate and fast object detection[C]// Proceedings of the 2018 European Conference on Computer Vision, LNCS 11215. Cham: Springer, 2018: 404-419. |
13 | 邵慧翔,曾丹. 基于改进YOLOv3算法的水下小目标分类与识别[J]. 上海大学学报(自然科学版), 2021, 27(3):481-491. 10.12066/j.issn.1007-2861.2279 |
SHAO H X, ZENG D. Classification and recognition of underwater small targets based on improved YOLOv3 algorithm[J]. Journal of Shanghai University (Natural Science Edition), 2021, 27(3):481-491. 10.12066/j.issn.1007-2861.2279 | |
14 | 于洋,李世杰,陈亮,等. 基于改进 YOLO v2 的船舶目标检测方法[J]. 计算机科学, 2019, 46(8): 332-336. |
YU Y, LI S J, CHEN L, et al. Ship target detection based on improved YOLO v2[J]. Computer Science, 2019, 46(8): 332-336. | |
15 | YE K Q, FANG Z B, HUANG X J, et al. Research on small target detection algorithm based on improved YOLOv3[C]// Proceedings of the 5th International Conference on Mechanical, Control and Computer Engineering. Piscataway: IEEE, 2020: 1467-1470. 10.1109/icmcce51767.2020.00321 |
16 | REZAEE M, ZHANG Y, MISHRA R, et al. Using a VGG-16 network for individual tree species detection with an object-based approach[C]// Proceedings of the 10th IAPR Workshop on Pattern Recognition in Remote Sensing. Piscataway: IEEE, 2018: 1-7. 10.1109/prrs.2018.8486395 |
17 | LI B Q, HE Y Y. An improved ResNet based on the adjustable shortcut connections[J]. IEEE Access, 2018, 6:18967-18974. 10.1109/access.2018.2814605 |
18 | TAN M X, PANG R M, LE Q V. EfficientDet: scalable and efficient object detection [C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 10778-10787. 10.1109/cvpr42600.2020.01079 |
19 | 王建军,魏江,梅少辉,等. 面向遥感图像小目标检测的改进YOLOv3算法[J]. 计算机工程与应用, 2021, 57(20): 133-141. |
WANG J J, WEI J, MEI S H, et al. Improved YOLOv3 for small target detection in remote sensing images[J]. Computer Engineering and Applications, 2021, 57(20): 133-141. |
[1] | 秦璟, 秦志光, 李发礼, 彭悦恒. 基于概率稀疏自注意力神经网络的重性抑郁疾患诊断[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2970-2974. |
[2] | 李力铤, 华蓓, 贺若舟, 徐况. 基于解耦注意力机制的多变量时序预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2732-2738. |
[3] | 赵志强, 马培红, 黑新宏. 基于双重注意力机制的人群计数方法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2886-2892. |
[4] | 薛凯鹏, 徐涛, 廖春节. 融合自监督和多层交叉注意力的多模态情感分析网络[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2387-2392. |
[5] | 汪雨晴, 朱广丽, 段文杰, 李书羽, 周若彤. 基于交互注意力机制的心理咨询文本情感分类模型[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2393-2399. |
[6] | 高鹏淇, 黄鹤鸣, 樊永红. 融合坐标与多头注意力机制的交互语音情感识别[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2400-2406. |
[7] | 李钟华, 白云起, 王雪津, 黄雷雷, 林初俊, 廖诗宇. 基于图像增强的低照度人脸检测[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2588-2594. |
[8] | 莫尚斌, 王文君, 董凌, 高盛祥, 余正涛. 基于多路信息聚合协同解码的单通道语音增强[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2611-2617. |
[9] | 李烨恒, 罗光圣, 苏前敏. 基于改进YOLOv5的Logo检测算法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2580-2587. |
[10] | 刘丽, 侯海金, 王安红, 张涛. 基于多尺度注意力的生成式信息隐藏算法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2102-2109. |
[11] | 徐松, 张文博, 王一帆. 基于时空信息的轻量视频显著性目标检测网络[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2192-2199. |
[12] | 李大海, 王忠华, 王振东. 结合空间域和频域信息的双分支低光照图像增强网络[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2175-2182. |
[13] | 魏文亮, 王阳萍, 岳彪, 王安政, 张哲. 基于光照权重分配和注意力的红外与可见光图像融合深度学习模型[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2183-2191. |
[14] | 熊武, 曹从军, 宋雪芳, 邵云龙, 王旭升. 基于多尺度混合域注意力机制的笔迹鉴别方法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2225-2232. |
[15] | 姬张建, 杜娜. 基于改进VariFocalNet的微小目标检测[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2200-2207. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||