ABSTRACT
Owing to the invisibility characteristics of the interiors of concrete structures, nondestructive testing technologies are commonly employed to detect internal damage. Electromagnetic flaw detection technology, as a prominent nondestructive testing method, offers capabilities of strong penetration and precise imaging. However, the output images produced by this detection technology are usually reliant on subjective human judgment, and existing automatic detection algorithms have limitations in terms of accuracy. To enhance the accuracy of electromagnetic flaw detection images, this paper proposes improved YOLOv7 based detection method. First of all, the ACBlock convolution module is introduced to replace the original 3×3 convolution kernel for alleviating the loss of edge feature information caused by the sampling operation of the backbone network MP module. Next, in order to have more focus on the semantic features of small targets, a global attention mechanism (GAMAttention) is incorporated into the algorithm. Then, the FReLU activation function is adopted to enhance the algorithm sensitivity to image space. Finally, experiments are conducted on a self-created concrete flaw detection image dataset to illustrate the efficiency of the proposed algorithm. The results show that the mean Average Precision (mAP) value achieved by the improved algorithm is 95.85%. which has a significant improvement of 13.89% in comparison with the pre-improved YOLOv7 algorithm and surpasses other algorithms in the same domain. These findings represent that the proposed algorithm effectively enhances the accuracy of concrete electromagnetic flaw detection image detection.
- Shan Zhao, and Imad L. Al-Qadi. 2016. Development of an analytic approach utilizing the extended common midpoint method to estimate asphalt pavement thickness with 3-D ground-penetrating radar. NDT & E International, 78(March 2016), 29-36. https://doi.org/10.1016/j.ndteint.2015.11.005.Google ScholarCross Ref
- Dar-Hao Chen, Feng Hong, Wujun Zhou, and Peng Ying. 2014. Estimating the hotmix asphalt air voids from ground penetrating radar. NDT & E International, 68(December 2014), 120-127. https://doi.org/10.1016/j.ndteint.2014.08.008.Google ScholarCross Ref
- Miguel A. Franesqui, Jorge Yepes, and Cándida García-González. 2017. Ultrasound data for laboratory calibration of an analytical model to calculate crack depth on asphalt pavements. Data in Brief, 13(July 2017), 723-730. https://doi.org/10.1016/j.dib.2017.06.053.Google ScholarCross Ref
- Shibin Lin, Dewei Meng, Hajin Choi, Sadegh Shams, and Hoda Azari. 2018. Laboratory assessment of nine methods for nondestructive evaluation of concrete bridge decks with overlays. Construction and Building Materials, 188(November 2018), 966-982. https://doi.org/10.1016/j.conbuildmat.2018.08.127.Google ScholarCross Ref
- Jiaming Tang, Chunhua Chen, Zhiyong Huang, Xiaoning Zhang, Weixiong Li, Min Huang, and Linghui Deng. 2022. Crack Unet: Crack Recognition Algorithm Based on Three-Dimensional Ground Penetrating Radar Images. Sensors 22, 23(December 2022): 9366. https://doi.org/10.3390/s22239366Google ScholarCross Ref
- M. -T. Pham and S. Lefèvre. 2018 Buried Object Detection from B-Scan Ground Penetrating Radar Data Using Faster-RCNN. IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium, IEEE, Valencia, Spain, 6804-6807. https://doi.org/10.1109/IGARSS.2018.8517683.Google ScholarCross Ref
- Wentai Lei, Feifei Hou, Jingchun Xi, Qianying Tan, Mengdi Xu, Xinyue Jiang, Gengye Liu, and Qingyuan Gu. 2019. Automatic hyperbola detection and fitting in GPR B-scan image. Automation in Construction, 106(October 2019). https://doi.org/10.1016/j.autcon.2019.102839.Google ScholarCross Ref
- Yu-Cheng Fan, Chun Ju Huang, and Chitra Meghala Yelamandala, 2023. Image Recognition Based on High Accuracy 3D Depth Map Information, Journal of Advances in Information Technology, 14, 5(2023), 1082-1087.Google ScholarCross Ref
- Hui Wang, Shan Ouyang, Qinghua Liu, Kefei Liao, and Lijun Zhou. 2022. Structure Feature Detection Method for Ground Penetrating Radar Two-Dimensional Profile Image Based on Deep Learning. Journal of Electronics & Information Technology, 44, 4 (2022), 1284-1294. https://doi.org/10.11999/JEIT211032.Google ScholarCross Ref
- Silva, Wilson Ricardo Leal da, and Diogo Schwerz de Lucena. 2018. Concrete Cracks Detection Based on Deep Learning Image Classification. Proceedings 2, 8(June 2018), 489. https://doi.org/10.3390/ICEM18-05387.Google ScholarCross Ref
- Shuwei Li, Xingyu Gu, Xiangrong Xu, Dawei Xu, Tianjie Zhang, Zhen Liu, and Qiao Dong. 2021. Detection of concealed cracks from ground penetrating radar images based on deep learning algorithm. Construction and Building Materials, 273(March 2021). https://doi.org/10.1016/j.conbuildmat.2020.121949Google ScholarCross Ref
- Chienyao Wang, Alexey Bochkovskiy, and Hongyuan Mark Liao. 2023. YOLOv7: Trainable Bag-of-Freebies Sets New State-of-the-Art for Real-Time Object Detectors. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, Vancouver, BC, Canada, 7464-7475, https://doi.org/10.1109/CVPR52729.2023.00721.Google ScholarCross Ref
- Bochkovskiy Alexey, Chien-Yao Wang, and Hong-Yuan Mark Liao. 2020. YOLOv4: Optimal Speed and Accuracy of Object Detection. arXiv, (2020), https://doi.org/10.48550/arXiv.2004.10934Google ScholarCross Ref
- Joseph Redmon, and Ali Farhadi. 2018. YOLOv3: An Incremental Improvement. arXiv. https://doi.org/10.48550/arXiv.1804.02767Google ScholarCross Ref
- Github 2021. Retrieved December 22, 2022 from https://github.com/ultralytics/yolov5.2021.Google Scholar
- Xiaohan Ding, Yuchen Guo, Guiguang Ding, and Jungong Han. 2019. ACNet: Strengthening the Kernel Skeletons for Powerful CNN via Asymmetric Convolution Blocks. International Conference on Computer Vision (ICCV), IEEE/CVF, Seoul, Korea (South), 1911-1920, https://doi.org/10.1109/ICCV.2019.00200.Google ScholarCross Ref
- Yichao Liu, Zongru Shao, and Nico Hoffmann. 2021. Global Attention Mechanism: Retain Information to Enhance Channel-Spatial Interactions. arXiv, (2021), https://doi.org/10.48550/arXiv.2112.05561.Google ScholarCross Ref
- Stefan Elfwing, Eiji Uchibe, and Kenji Doya. 2018. Sigmoid weight-ed linear unitsfor neural network function approximation in reinforcement learning. Neural Networks, 107(2018), 3-11. https://doi.org/10.48550/arXiv.1702.03118.Google ScholarCross Ref
- Suo Qiu, Xiangmin Xu, and Bolun Cai. 2018. FReLU: Flexible Rectified Linear Units for Improving Convolutional Neural Networks. 2018 24th International Conference on Pattern Recognition (ICPR), IEEE, Beijing, China, 1223-1228. c10.1109/ICPR.2018.8546022.Google Scholar
- Jinyang Zhou, and Ming Chen. 2023. Ground-penetrating radar earth and rock embankment cavity detection method based on improved YOLOv7. Foreign Electronic Measurement Technology, 42,07(2023), 146-153. https://doi.org/10.3969/j.issn.1002-8978.2023.07.020.Google ScholarCross Ref
- Ming Xu, Gong Zhang, Guangtao Wang, Geng Luo, and Ruibo Zheng. 2023. Research and application of anomaly detection methods in ground-penetr-ating radar images. Coal Geology of China, 35, 2 (2023), 73-78.Google Scholar
- Pengfei Lai, Wei Li, Yao Gao, Jiangang Ding, Bo Yuan, and Ming Yang. 2023. GPR Pipeline Target Detection Based on Improved Cascade R-CNN. Computer Systems&Applications, 32, 2, (2023), 102-110. https://doi.org/10.15888/j.cnki.csa.008945.Google ScholarCross Ref
Index Terms
- Image Recognition of Concrete Electromagnetic Flaw Detection Based on Improved YOLOv7
Recommendations
Helmet detection algorithm based on improved yolov7
CSAI '23: Proceedings of the 2023 7th International Conference on Computer Science and Artificial IntelligenceIn order to solve the problems of low detection accuracy, wrong detection and missed detection in the traditional target detection algorithm in the safety helmet detection task, a safety helmet detection algorithm based on improved YOLOv7 is proposed. ...
Driver fatigue detection based on improved YOLOv7
AbstractFatigue driving is one of the main reasons threatening road traffic safety. Aiming at the problems of complex detection process, low accuracy, and susceptibility to light interference in the current driver fatigue detection algorithm, this paper ...
Improved YOLOv7 UAV image small target detection algorithm
CSAI '23: Proceedings of the 2023 7th International Conference on Computer Science and Artificial IntelligenceIn order to improve the detection efficiency of small targets in UAV images in traffic systems, a small target detection algorithm for UAV images in traffic systems based on improved YOLOv7 is proposed. In the feature fusion network, the small target ...
Comments