A Lightweight and High-Accuracy Model for Pavement Crack Segmentation DOI Creative Commons

Yuhui Yu,

Wenjun Xia,

Zhangyan Zhao

и другие.

Applied Sciences, Год журнала: 2024, Номер 14(24), С. 11632 - 11632

Опубликована: Дек. 12, 2024

Pavement cracks significantly affect road safety and longevity, making accurate crack segmentation essential for effective maintenance. Although deep learning methods have demonstrated excellent performance in this task, their large network architectures limit applicability on resource-constrained devices. To address challenge, paper proposes a lightweight, fully convolutional neural model, enhanced with spatial information. First, the backbone structure is optimized to improve efficiency of information utilization. Second, by incorporating adaptive feature reassembly wavelet transforms, up-sampling down-sampling processes are refined, enhancing model capacity capture Lastly, dynamic combined loss function employed during training further attention edge details. validate performance, we trained tested it Crack500 dataset applied directly AsphaltCrack300 dataset. Experimental results indicate that proposed achieved an MIoU 80.37% F1-score 78.22% dataset, representing increases 3.08% 5.62%, respectively, compared EfficientNet. On exhibited strong robustness, outperforming other mainstream models. Additionally, its lightweight design provides clear advantages, well suited realworld applications limited computational resources.

Язык: Английский

FTN-ResNet50: flexible transformer network model with ResNet50 for road crack detection DOI
Y. H. Lin,

Tao Yu,

Zheshuai Lin

и другие.

Evolving Systems, Год журнала: 2025, Номер 16(2)

Опубликована: Март 24, 2025

Язык: Английский

Процитировано

0
A Novel YOLOv10-DECA Model for Real-Time Detection of Concrete Cracks DOI Creative Commons

Chaokai Zhang,

Ningbo Peng, Jiaheng Yan

и другие.

Buildings, Год журнала: 2024, Номер 14(10), С. 3230 - 3230

Опубликована: Окт. 11, 2024

The You Only Look Once (YOLO) series algorithms have been widely adopted in concrete crack detection, with attention mechanisms frequently being incorporated to enhance recognition accuracy and efficiency. However, existing research is confronted by two primary challenges: the suboptimal performance of mechanism modules lack explanation regarding how these influence model’s decision-making process improve accuracy. To address issues, a novel Dynamic Efficient Channel Attention (DECA) module proposed this study, which designed YOLOv10 model effectiveness visually demonstrated through application interpretable analysis algorithms. In paper, dataset complex background used. Experimental results indicate that DECA significantly improves localization detection discontinuous cracks, outperforming (ECA). When compared similarly sized YOLOv10n model, YOLOv10-DECA demonstrates improvements 4.40%, 3.06%, 4.48%, 5.56% precision, recall, mAP50, mAP50-95 metrics, respectively. Moreover, even when larger YOLOv10s indicators are increased 2.00%, 0.04%, 2.27%, 1.12%, terms speed evaluation, owing lightweight design module, achieves an inference 78 frames per second, 2.5 times faster than YOLOv10s, thereby fully meeting requirements for real-time detection. These demonstrate optimized balance between tasks has achieved model. Consequently, study provides valuable insights future applications field.

Язык: Английский

Процитировано

3
LANA-YOLO: Road defect detection algorithm optimized for embedded solutions DOI Creative Commons
Paweł Tomiło

Applied Computer Science, Год журнала: 2025, Номер 21(1), С. 164 - 181

Опубликована: Март 31, 2025

Poor pavement condition leads to increased risk of accidents, vehicle damage, and reduced transportation efficiency. The author points out that traditional methods monitoring road conditions are time-consuming costly, so a modern approach based on the use developed neural network model is presented. main aim this paper create can infer in real time, with less computing power maintaining or improving metrics base model, YOLOv8. Based assumption, architecture LANA-YOLOv8 (Large Kernel Attention Involution Asymptotic Feature Pyramid) proposed. model's tailored operate environments limited resources, including single-board minicomputers. In addition, article presents Basic Block (BIB) uses involution layer provide better performance at lower cost than convolution layers. was compared other architectures public dataset as well specially created for these purposes. solution has requirements, which translates into faster inference times. At same achieved results validation tests against model.

Язык: Английский

Процитировано

0
CV-YOLOv10-AR-M: Foreign Object Detection in Pu-Erh Tea Based on Five-Fold Cross-Validation DOI Creative Commons
Wenxia Yuan, Chunhua Yang, Xinghua Wang

и другие.

Foods, Год журнала: 2025, Номер 14(10), С. 1680 - 1680

Опубликована: Май 9, 2025

To address the problem of detecting foreign bodies in Pu-erh tea, this study proposes an intelligent detection method based on improved YOLOv10 network. By introducing MPDIoU loss function, network is optimized to effectively enhance positioning accuracy model complex background and improve small target objects. Using AssemFormer optimize structure, network’s ability perceive objects its process global information are improved. Rectangular Self-Calibrated Module, prediction bounding box optimized, further improving classification target-positioning abilities scenes. The results showed that Box, Cls, Dfl functions CV-YOLOv10-AR-M One-to-Many Head task were, respectively, 14.60%, 19.74%, 20.15% lower than those In One-to-One task, they decreased by 10.42%, 29.11%, 20.15%, respectively. Compared with original network, accuracy, recall rate, mAP were increased 5.35%, 11.72% 8.32%, improves model’s attention sizes, backgrounds, detailed information, providing effective technical support for quality control agricultural field.

Язык: Английский

Процитировано

0
Survey of automated crack detection methods for asphalt and concrete structures DOI
Oumaima Khlifati, Khadija Baba, Bassam A. Tayeh

и другие.

Innovative Infrastructure Solutions, Год журнала: 2024, Номер 9(11)

Опубликована: Окт. 27, 2024

Язык: Английский

Процитировано

3
TS-GRU: A Stock Gated Recurrent Unit Model Driven via Neuro-Inspired Computation DOI Open Access
Yuanfang Zhang, Heinz Fill

Electronics, Год журнала: 2024, Номер 13(23), С. 4659 - 4659

Опубликована: Ноя. 26, 2024

Existing risk measurement methods often fail to fully consider the impact of climatic conditions on stock market risk, making it difficult capture dynamic patterns and long-term dependencies. To address these issues, we propose TS-GRU method: this approach utilizes a temporal convolutional network (TCN) extract underlying features from historical data, capturing key characteristics time series data. Subsequently, gated recurrent unit (GRU) model is employed dependencies within market. Finally, optimized using Sparrow algorithm based collective behavior, iteratively evaluating refining parameters obtain improved solutions. Experimental results demonstrate effectiveness method in providing accurate assessment forecasting. This comprehensive takes into account carbon finance, climate change, environmental factors, offering valuable insights investors help them understand manage investment risks ever-changing

Язык: Английский

Процитировано

1
Mixed Reality-Based Concrete Crack Detection and Skeleton Extraction Using Deep Learning and Image Processing DOI Open Access
Davood Shojaei, Peyman Jafary, Zezheng Zhang

и другие.

Electronics, Год журнала: 2024, Номер 13(22), С. 4426 - 4426

Опубликована: Ноя. 12, 2024

Advancements in image processing and deep learning offer considerable opportunities for automated defect assessment civil structures. However, these systems cannot work interactively with human inspectors. Mixed reality (MR) can be adopted to address this by involving inspectors various stages of the process. This paper integrates You Only Look Once (YOLO) v5n YOLO v5m Canny algorithm real-time concrete crack detection skeleton extraction a Microsoft HoloLens 2 MR device. The demonstrates superior mean average precision (mAP) 0.5 speed, while achieves highest mAP 0.95 among other v5 also outperforms Sobel Prewitt edge detectors F1 score. developed MR-based system could not only employed but utilized automatic recording location specifications cracks further analysis future re-inspections.

Язык: Английский

Процитировано

0
A Lightweight and High-Accuracy Model for Pavement Crack Segmentation DOI Creative Commons

Yuhui Yu,

Wenjun Xia,

Zhangyan Zhao

и другие.

Applied Sciences, Год журнала: 2024, Номер 14(24), С. 11632 - 11632

Опубликована: Дек. 12, 2024

Pavement cracks significantly affect road safety and longevity, making accurate crack segmentation essential for effective maintenance. Although deep learning methods have demonstrated excellent performance in this task, their large network architectures limit applicability on resource-constrained devices. To address challenge, paper proposes a lightweight, fully convolutional neural model, enhanced with spatial information. First, the backbone structure is optimized to improve efficiency of information utilization. Second, by incorporating adaptive feature reassembly wavelet transforms, up-sampling down-sampling processes are refined, enhancing model capacity capture Lastly, dynamic combined loss function employed during training further attention edge details. validate performance, we trained tested it Crack500 dataset applied directly AsphaltCrack300 dataset. Experimental results indicate that proposed achieved an MIoU 80.37% F1-score 78.22% dataset, representing increases 3.08% 5.62%, respectively, compared EfficientNet. On exhibited strong robustness, outperforming other mainstream models. Additionally, its lightweight design provides clear advantages, well suited realworld applications limited computational resources.

Язык: Английский

Процитировано

0