《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (1): 265-272.DOI: 10.11772/j.issn.1001-9081.2021010206
• 前沿与综合应用 • 上一篇
许慧青1,2, 陈斌2(), 王敬飞3,4, 陈志毅3,4, 覃健5
收稿日期:
2021-02-03
修回日期:
2021-04-25
接受日期:
2021-04-28
发布日期:
2022-01-11
出版日期:
2022-01-10
通讯作者:
陈斌
作者简介:
许慧青(1989—),男,河南汝南人,博士研究生,主要研究方向:目标检测、语义分割、路面病害智能检测Huiqing XU1,2, Bin CHEN2(), Jingfei WANG3,4, Zhiyi CHEN3,4, Jian QIN5
Received:
2021-02-03
Revised:
2021-04-25
Accepted:
2021-04-28
Online:
2022-01-11
Published:
2022-01-10
Contact:
Bin CHEN
About author:
XU Huiqing, born in 1989, Ph. D. candidate. His research interests include target detection, semantic segmentation, intelligent detection of pavement distress.摘要:
针对细长路面病害人工检测耗时长和当前检测方法精度不足的问题,依据病害的弱语义特性和异常几何属性,提出了能够精准定位和分类出病害的二阶段细长路面病害检测方法Epd RCNN。首先,针对细长路面病害的弱语义特性,提出了一种复用低层特征并反复融合不同阶段特征的骨干网络;其次,在训练过程中,使用一种符合病害几何属性分布的锚框机制来生成高质量的正样本供网络训练;然后,在单一高分辨率特征图上预测病害包围框,并针对该特征图使用并行级联空洞卷积模块来提升其多尺度特征表达能力;最后,针对形状各异的候选区域,使用由可变形感兴趣区域池化(RoI Pooling)和空间注意力模块组成的候选区域特征改良模块来提取符合病害几何属性的候选区域特征。实验结果表明,所提方法在光照充足图像上的平均准确率均值(mAP)为0.907,在存在光照问题图像上的mAP为0.891,综合mAP为0.899, 表明该方法具有良好的检测性能和对光照的鲁棒性。
中图分类号:
许慧青, 陈斌, 王敬飞, 陈志毅, 覃健. 基于卷积神经网络的细长路面病害检测方法[J]. 计算机应用, 2022, 42(1): 265-272.
Huiqing XU, Bin CHEN, Jingfei WANG, Zhiyi CHEN, Jian QIN. Elongated pavement distress detection method based on convolutional neural network[J]. Journal of Computer Applications, 2022, 42(1): 265-272.
类别 | 训练集图像数 | 测试集图像数 | 图像的总数 |
---|---|---|---|
纵向裂缝 | 1 866 | 207 | 2 073 |
横向裂缝 | 1 901 | 211 | 2 112 |
修补 | 1 923 | 213 | 2 136 |
修补不良 | 1 911 | 212 | 2 123 |
表1 训练集和测试集中的图像数
Tab. 1 Number of images in training set and test set
类别 | 训练集图像数 | 测试集图像数 | 图像的总数 |
---|---|---|---|
纵向裂缝 | 1 866 | 207 | 2 073 |
横向裂缝 | 1 901 | 211 | 2 112 |
修补 | 1 923 | 213 | 2 136 |
修补不良 | 1 911 | 212 | 2 123 |
锚框机制 | 尺度 | 长宽比 |
---|---|---|
FPN | [ | [0.5, 1.0, 2.0] |
MSAM | [ | [0.04, 0.1, 0.5, 1.0, 2.0, 10.0, 25.0] |
表2 训练过程中使用的两种锚框机制
Tab. 2 Two anchor box mechanisms used in training process
锚框机制 | 尺度 | 长宽比 |
---|---|---|
FPN | [ | [0.5, 1.0, 2.0] |
MSAM | [ | [0.04, 0.1, 0.5, 1.0, 2.0, 10.0, 25.0] |
检测方法 | 骨干网络 |
---|---|
Faster RCNN | ResNet101 |
Faster RCNN+ | VovNetV2-57 |
Epd RCNN- | HRNETV2-W32 |
Epd RCNN | HRNet++ |
表3 对比实验中的骨干网络和检测方法
Tab. 3 Backbone networks and detection methods in comparison experiments
检测方法 | 骨干网络 |
---|---|
Faster RCNN | ResNet101 |
Faster RCNN+ | VovNetV2-57 |
Epd RCNN- | HRNETV2-W32 |
Epd RCNN | HRNet++ |
检测方法 | 骨干网络 | 平均精度 | mAP | mAP_B | mAP_D | |||
---|---|---|---|---|---|---|---|---|
纵向裂缝 | 横向裂缝 | 修补 | 修补不良 | |||||
Faster RCNN | ResNet101 | 0.423 | 0.472 | 0.663 | 0.499 | 0.514 | 0.519 | 0.510 |
Faster RCNN + | 0.456 | 0.495 | 0.683 | 0.534 | 0.542 | 0.551 | 0.533 | |
Epd RCNN- | 0.472 | 0.497 | 0.689 | 0.552 | 0.552 | 0.557 | 0.548 | |
Epd RCNN | 0.516 | 0.524 | 0.723 | 0.608 | 0.592 | 0.598 | 0.587 | |
Faster RCNN | VoVNetV2-57 | 0.435 | 0.488 | 0.672 | 0.512 | 0.526 | 0.531 | 0.523 |
Faster RCNN + | 0.446 | 0.492 | 0.692 | 0.523 | 0.538 | 0.542 | 0.534 | |
Epd RCNN- | 0.491 | 0.522 | 0.701 | 0.543 | 0.564 | 0.568 | 0.561 | |
Epd RCNN | 0.526 | 0.525 | 0.715 | 0.614 | 0.595 | 0.603 | 0.587 | |
Faster RCNN | HRNetV2-W32 | 0.549 | 0.564 | 0.765 | 0.659 | 0.634 | 0.641 | 0.616 |
Faster RCNN + | 0.556 | 0.596 | 0.779 | 0.662 | 0.648 | 0.651 | 0.646 | |
Epd RCNN- | 0.637 | 0.655 | 0.804 | 0.728 | 0.706 | 0.711 | 0.701 | |
Epd RCNN | 0.686 | 0.683 | 0.809 | 0.774 | 0.738 | 0.742 | 0.734 | |
Faster RCNN | HRNet++ | 0.625 | 0.718 | 0.826 | 0.774 | 0.736 | 0.739 | 0.733 |
Faster RCNN + | 0.703 | 0.732 | 0.846 | 0.792 | 0.768 | 0.771 | 0.766 | |
Epd RCNN- | 0.773 | 0.788 | 0.863 | 0.818 | 0.811 | 0.820 | 0.801 | |
Epd RCNN | 0.893 | 0.895 | 0.912 | 0.897 | 0.899 | 0.907 | 0.891 |
表4 在测试集上细长路面病害的检测性能比较
Tab. 4 Comparison of elongated pavement distress detection performance on test set
检测方法 | 骨干网络 | 平均精度 | mAP | mAP_B | mAP_D | |||
---|---|---|---|---|---|---|---|---|
纵向裂缝 | 横向裂缝 | 修补 | 修补不良 | |||||
Faster RCNN | ResNet101 | 0.423 | 0.472 | 0.663 | 0.499 | 0.514 | 0.519 | 0.510 |
Faster RCNN + | 0.456 | 0.495 | 0.683 | 0.534 | 0.542 | 0.551 | 0.533 | |
Epd RCNN- | 0.472 | 0.497 | 0.689 | 0.552 | 0.552 | 0.557 | 0.548 | |
Epd RCNN | 0.516 | 0.524 | 0.723 | 0.608 | 0.592 | 0.598 | 0.587 | |
Faster RCNN | VoVNetV2-57 | 0.435 | 0.488 | 0.672 | 0.512 | 0.526 | 0.531 | 0.523 |
Faster RCNN + | 0.446 | 0.492 | 0.692 | 0.523 | 0.538 | 0.542 | 0.534 | |
Epd RCNN- | 0.491 | 0.522 | 0.701 | 0.543 | 0.564 | 0.568 | 0.561 | |
Epd RCNN | 0.526 | 0.525 | 0.715 | 0.614 | 0.595 | 0.603 | 0.587 | |
Faster RCNN | HRNetV2-W32 | 0.549 | 0.564 | 0.765 | 0.659 | 0.634 | 0.641 | 0.616 |
Faster RCNN + | 0.556 | 0.596 | 0.779 | 0.662 | 0.648 | 0.651 | 0.646 | |
Epd RCNN- | 0.637 | 0.655 | 0.804 | 0.728 | 0.706 | 0.711 | 0.701 | |
Epd RCNN | 0.686 | 0.683 | 0.809 | 0.774 | 0.738 | 0.742 | 0.734 | |
Faster RCNN | HRNet++ | 0.625 | 0.718 | 0.826 | 0.774 | 0.736 | 0.739 | 0.733 |
Faster RCNN + | 0.703 | 0.732 | 0.846 | 0.792 | 0.768 | 0.771 | 0.766 | |
Epd RCNN- | 0.773 | 0.788 | 0.863 | 0.818 | 0.811 | 0.820 | 0.801 | |
Epd RCNN | 0.893 | 0.895 | 0.912 | 0.897 | 0.899 | 0.907 | 0.891 |
图7 Epd RCNN在不同光照条件下正确的细长路面病害检测结果以及该方法错误的检测结果
Fig. 7 Correct elongated pavement distress detection results of Epd RCNN under different illumination conditions and itswrong detection results
1 | LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]// Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2017: 936-944. 10.1109/cvpr.2017.106 |
2 | LI Z M, PENG C, YU G, et al. DetNet: design backbone for object detection[C]// Proceedings of the 2018 European Conference on Computer Vision, LNCS11213. Cham: Springer, 2018: 339-354. |
3 | KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]// Proceedings of the 25th International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2012: 1097-1105. |
4 | 初秀民,王荣本,储江伟,等. 基于不变矩特征的沥青路面破损图像识别[J]. 吉林大学学报(工学版), 2003, 33(1): 1-7. 10.3969/j.issn.1671-5497.2003.01.001 |
CHU X M, WANG R B, CHU J W, et al. Asphalt pavement surface distress image recognition based on moment invariant feature[J]. Journal of Jilin University (Engineering and Technology Edition), 2003, 33(1): 1-7. 10.3969/j.issn.1671-5497.2003.01.001 | |
5 | AYENU-PRAH A, ATTOH-OKINE N. Evaluating pavement cracks with bidimensional empirical mode decomposition[J]. EURASIP Journal on Advances in Signal Processing, 2008, 2008: No.861701. 10.1155/2008/861701 |
6 | DIXIT A, WAGATSUMA H. Investigating the effectiveness of the Sobel operator in the MCA-based automatic crack detection[C]// Proceedings of the 4th International Conference on Optimization and Applications. Piscataway: IEEE, 2018: 1-6. 10.1109/icoa.2018.8370555 |
7 | TANG J S, GU Y L. Automatic crack detection and segmentation using a hybrid algorithm for road distress analysis[C]// Proceedings of the 2013 IEEE International Conference on Systems, Man, and Cybernetics. Piscataway: IEEE, 2013: 3026-3030. 10.1109/smc.2013.516 |
8 | LI Q Q, LIU X L. Novel approach to pavement image segmentation based on neighboring difference histogram method[C]// Proceedings of the 2008 Congress on Image and Signal Processing. Piscataway: IEEE, 2008: 792-796. 10.1109/CISP.2008.13 |
9 | OLIVEIRA H, CORREIA P L. Automatic road crack segmentation using entropy and image dynamic thresholding[C]// Proceedings of the 17th European Signal Processing Conference. Piscataway: IEEE, 2009: 622-626. 10.1109/icip.2010.5653305 |
10 | CHENG H D, MIYOJIM M. Automatic pavement distress detection system[J]. Information Sciences, 1998, 108(1/2/3/4): 219-240. 10.1016/s0020-0255(97)10062-7 |
11 | HU Y, ZHAO C X. A novel local binary pattern based methods for pavement crack detection[J]. Journal of Pattern Recognition Research, 2010, 5(1): 140-147. 10.13176/11.167 |
12 | SALMAN M, MATHAVAN S, KAMAL K, et al. Pavement crack detection using the Gabor filter[C]// Proceedings of the 16th International IEEE Conference on Intelligent Transportation Systems. Piscataway: IEEE, 2013: 2039-2044. 10.1109/itsc.2013.6728529 |
13 | KAPELA R, ŚNIATAŁA P, TURKOT A, et al. Asphalt surfaced pavement cracks detection based on histograms of oriented gradients[C]// Proceedings of the 22nd International Conference “Mixed Design of Integrated Circuits and Systems”. Piscataway: IEEE, 2015: 579-584. 10.1109/mixdes.2015.7208590 |
14 | HADJI I, WILDES R P. What do we understand about convolutional networks ?[EB/OL]. (2018-03-23) [2020-03-01].. |
15 | 沙爱民,童峥,高杰. 基于卷积神经网络的路表病害识别与测量[J]. 中国路面学报, 2018, 31(1): 1-10. |
SHA A M, TONG Z, GAO J. Recognition and measurement of pavement disasters based on convolutional neural networks[J]. China Journal of Highway and Transport, 2018, 31(1): 1-10. | |
16 | ZHANG L, YANG F, ZHANG Y D, et al. Road crack detection using deep convolutional neural network[C]// Proceedings of the 2016 IEEE International Conference on Image Processing. Piscataway: IEEE, 2016: 3708-3712. 10.1109/icip.2016.7533052 |
17 | WANG X L, HU Z Z. Grid-based pavement crack analysis using deep learning[C]// Proceedings of the 4th International Conference on Transportation Information and Safety. Piscataway: IEEE, 2017: 917-924. 10.1109/ictis.2017.8047878 |
18 | YANG X C, LI H, YU Y T, et al. Automatic pixel-level crack detection and measurement using fully convolutional network[J]. Computer-Aided Civil and Infrastructure Engineering, 2018, 33(12): 1090-1109. 10.1111/mice.12412 |
19 | HUANG H W, LI Q T, ZHANG D M. Deep learning based image recognition for crack and leakage defects of metro shield tunnel[J]. Tunnelling and Underground Space Technology, 2018, 77: 166-176. 10.1016/j.tust.2018.04.002 |
20 | FAN R, BOCUS M J, ZHU Y L, et al. Road crack detection using deep convolutional neural network and adaptive thresholding[C]// Proceedings of the 2019 IEEE Intelligent Vehicles Symposium. Piscataway: IEEE, 2019: 474-479. 10.1109/ivs.2019.8814000 |
21 | SHELHAMER E, LONG J, DARRELL T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4): 640-651. 10.1109/tpami.2016.2572683 |
22 | MAEDA H, SEKIMOTO Y, SETO T, et al. Road damage detection and classification using deep neural networks with smartphone images[J]. Computer-Aided Civil and Infrastructure Engineering, 2018, 33(12): 1127-1141. 10.1111/mice.12387 |
23 | WANG W Z, WU B, YANG S X, et al. Road damage detection and classification with Faster R-CNN[C]// Proceedings of the 2018 IEEE International Conference on Big Data. Piscataway: IEEE, 2018: 5220-5223. 10.1109/bigdata.2018.8622354 |
24 | REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149. 10.1109/tpami.2016.2577031 |
25 | 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 |
26 | ANAND S, GUPTA S, DARBARI V, et al. Crack-pot: autonomous road crack and pothole detection[C]// Proceedings of the 2018 Digital Image Computing: Techniques and Applications. Piscataway: IEEE, 2018: 1-6. 10.1109/dicta.2018.8615819 |
27 | ZHU X Z, HU H, LIN S, et al. Deformable ConvNets V2: more deformable, better results[C]// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2019: 9300-9308. 10.1109/cvpr.2019.00953 |
28 | WANG J D, SUN K, CHENG T H, et al. Deep high-resolution representation learning for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(10): 3349-3364. 10.1109/tpami.2020.2983686 |
29 | HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 770-778. 10.1109/cvpr.2016.90 |
30 | BUSLAEV A, IGLOVIKOV V I, KHVEDCHENYA E, et al. Albumentations: fast and flexible image augmentations[J]. Information, 2020, 11(2): No.125. 10.3390/info11020125 |
31 | EVERINGHAM M, GOOL L VAN, WILLIAMS C K I, et al. The PASCAL Visual Object Classes (VOC) challenge[J]. International Journal of Computer Vision, 2010, 88(2): 303-338. 10.1007/s11263-009-0275-4 |
32 | CHEN K, PANG J Q, WANG J M, et al. MMDetection: open MMLab detection toolbox and benchmark[EB/OL]. (2019-06-17) [2020-07-12].. |
33 | LIN T Y, MAIRE M, BELONGIE S, et al. Microsoft COCO: common objects in context[C]// Proceedings of the 2014 European Conference on Computer Vision, LNCS8693. Cham: Springer, 2014: 740-755. |
34 | LEE Y, PARK J. CenterMask: real-time anchor-free instance segmentation[C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 13903-13912. 10.1109/cvpr42600.2020.01392 |
[1] | 包银鑫, 曹阳, 施佺. 基于改进时空残差卷积神经网络的城市路网短时交通流预测[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 258-264. |
[2] | 王素玉, 杨静, 李越. 基于双注意力机制信息蒸馏网络的图像超分辨率复原算法[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 239-244. |
[3] | 李恒鑫, 常侃, 谭宇飞, 凌铭阳, 覃团发. 应用通道间相关性及增强信息蒸馏的彩色图像去马赛克网络[J]. 《计算机应用》唯一官方网站, 2022, 42(1): 245-251. |
[4] | 王贺兵, 张春梅. 基于非对称卷积-压缩激发-次代残差网络的人脸关键点检测[J]. 计算机应用, 2021, 41(9): 2741-2747. |
[5] | 宋中山, 梁家锐, 郑禄, 刘振宇, 帖军. 基于双向门控尺度特征融合的遥感场景分类[J]. 计算机应用, 2021, 41(9): 2726-2735. |
[6] | 李康康, 张静. 基于注意力机制的多层次编码和解码的图像描述模型[J]. 计算机应用, 2021, 41(9): 2504-2509. |
[7] | 张永斌, 常文欣, 孙连山, 张航. 基于字典的域名生成算法生成域名的检测方法[J]. 计算机应用, 2021, 41(9): 2609-2614. |
[8] | 赵宏, 孔东一. 图像特征注意力与自适应注意力融合的图像内容中文描述[J]. 计算机应用, 2021, 41(9): 2496-2503. |
[9] | 徐江浪, 李林燕, 万新军, 胡伏原. 结合目标检测的室内场景识别方法[J]. 计算机应用, 2021, 41(9): 2720-2725. |
[10] | 牟长宁, 王海鹏, 周丕宇, 侯鑫行. 基于图卷积神经网络的串联质谱从头测序[J]. 计算机应用, 2021, 41(9): 2773-2779. |
[11] | 曾祥银, 郑伯川, 刘丹. 基于深度卷积神经网络和聚类的左右轨道线检测[J]. 计算机应用, 2021, 41(8): 2324-2329. |
[12] | 曹玉红, 徐海, 刘荪傲, 王紫霄, 李宏亮. 基于深度学习的医学影像分割研究综述[J]. 《计算机应用》唯一官方网站, 2021, 41(8): 2273-2287. |
[13] | 秦斌斌, 彭良康, 卢向明, 钱江波. 司机分心驾驶检测研究进展[J]. 计算机应用, 2021, 41(8): 2330-2337. |
[14] | 黄程程, 董霄霄, 李钊. 基于二维Winograd算法的深流水线5×5卷积方法[J]. 计算机应用, 2021, 41(8): 2258-2264. |
[15] | 高钦泉, 黄炳城, 刘文哲, 童同. 基于改进CenterNet的竹条表面缺陷检测方法[J]. 计算机应用, 2021, 41(7): 1933-1938. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||