Hybrid Machine Learning for Automated Road Safety Inspection of Auckland Harbour Bridge DOI Open Access
Munish Rathee, Boris Bačić, Maryam Doborjeh

и другие.

Electronics, Год журнала: 2024, Номер 13(15), С. 3030 - 3030

Опубликована: Авг. 1, 2024

The Auckland Harbour Bridge (AHB) utilises a movable concrete barrier (MCB) to regulate the uneven bidirectional flow of daily traffic. In addition risk human error during regular visual inspections, staff members inspecting MCB work in diverse weather and light conditions, exerting themselves ergonomically unhealthy inspection postures with added weight protection gear mitigate risks, e.g., flying debris. To augment inspections an using computer vision technology, this study introduces hybrid deep learning solution that combines kernel manipulation custom transfer strategies. video data recordings were captured conditions (under safety supervision industry experts) involving high-speed (120 fps) camera system attached vehicle. Before identifying hazard, unsafe position pin connecting two 750 kg segments MCB, multi-stage preprocessing spatiotemporal region interest (ROI) involves rolling window before frames containing diagnostic information. This ResNet-50 architecture, enhanced 3D convolutions, within STENet framework capture analyse data, facilitating real-time surveillance (AHB). Considering sparse nature anomalies, initial peer-reviewed binary classification results (82.6%) for safe (intervention-required) scenarios improved 93.6% by incorporating synthetic expert feedback, retraining model. adaptation allowed optimised detection false positives negatives. future, we aim extend anomaly methods various infrastructure enhancing urban resilience, transport efficiency safety.

Язык: Английский

YOLO-RRL: A Lightweight Algorithm for PCB Surface Defect Detection DOI Creative Commons
Tian Zhang, Jie Zhang,

Pengfei Pan

и другие.

Applied Sciences, Год журнала: 2024, Номер 14(17), С. 7460 - 7460

Опубликована: Авг. 23, 2024

Printed circuit boards present several challenges to the detection of defects, including targets insufficient size and distribution, a high level background noise, variety complex types. These factors contribute difficulties encountered by PCB defect networks in accurately identifying defects. This paper proposes less-parametric model, YOLO-RRL, based on improved YOLOv8 architecture. The YOLO-RRL model incorporates four key improvement modules: following modules have been incorporated into proposed model: Robust Feature Downsampling (RFD), Reparameterised Generalised FPN (RepGFPN), Dynamic Upsampler (DySample), Lightweight Asymmetric Detection Head (LADH-Head). results multiple performance metrics evaluation demonstrate that enhances mean accuracy (mAP) 2.2 percentage points 95.2%, increases frame rate (FPS) 12%, significantly reduces number parameters computational complexity, thereby achieving balance between efficiency. Two datasets, NEU-DET APSPC, were employed evaluate YOLO-RRL. indicate exhibits good adaptability. In comparison existing mainstream inspection models, is also more advanced. capable improving production quality reducing costs practical applications while extending scope system wide range industrial applications.

Язык: Английский

Процитировано

3
An efficient and real-time steel surface defect detection method based on single-stage detection algorithm DOI
Hongkai Zhang,

Qiqi Miao,

Suqiang Li

и другие.

Multimedia Tools and Applications, Год журнала: 2024, Номер unknown

Опубликована: Июнь 15, 2024

Язык: Английский

Процитировано

1
Few-shot anomaly detection using positive unlabeled learning with cycle consistency and co-occurrence features DOI
Sion An, J. H. Kim, Soopil Kim

и другие.

Expert Systems with Applications, Год журнала: 2024, Номер 256, С. 124890 - 124890

Опубликована: Июль 27, 2024

Язык: Английский

Процитировано

1
A multimodal data generation method for imbalanced classification with dual-discriminator constrained diffusion model and adaptive sample selection strategy DOI

Qiangwei Li,

Xin Gao,

Heping Lu

и другие.

Information Fusion, Год журнала: 2024, Номер 117, С. 102843 - 102843

Опубликована: Дек. 5, 2024

Язык: Английский

Процитировано

1
Hybrid Machine Learning for Automated Road Safety Inspection of Auckland Harbour Bridge DOI Open Access
Munish Rathee, Boris Bačić, Maryam Doborjeh

и другие.

Electronics, Год журнала: 2024, Номер 13(15), С. 3030 - 3030

Опубликована: Авг. 1, 2024

The Auckland Harbour Bridge (AHB) utilises a movable concrete barrier (MCB) to regulate the uneven bidirectional flow of daily traffic. In addition risk human error during regular visual inspections, staff members inspecting MCB work in diverse weather and light conditions, exerting themselves ergonomically unhealthy inspection postures with added weight protection gear mitigate risks, e.g., flying debris. To augment inspections an using computer vision technology, this study introduces hybrid deep learning solution that combines kernel manipulation custom transfer strategies. video data recordings were captured conditions (under safety supervision industry experts) involving high-speed (120 fps) camera system attached vehicle. Before identifying hazard, unsafe position pin connecting two 750 kg segments MCB, multi-stage preprocessing spatiotemporal region interest (ROI) involves rolling window before frames containing diagnostic information. This ResNet-50 architecture, enhanced 3D convolutions, within STENet framework capture analyse data, facilitating real-time surveillance (AHB). Considering sparse nature anomalies, initial peer-reviewed binary classification results (82.6%) for safe (intervention-required) scenarios improved 93.6% by incorporating synthetic expert feedback, retraining model. adaptation allowed optimised detection false positives negatives. future, we aim extend anomaly methods various infrastructure enhancing urban resilience, transport efficiency safety.

Язык: Английский

Процитировано

0