At present, the method of two-dimensional image recognition is mainly used to detect the abnormal fastener in the rail-track inspection system. However, the too-tight-or-too-loose fastener condition may cause the clip of the fastener to break or loose due to the high frequency vibration shock, which is difficult to detect from the two-dimensional image. In this practical application background, 3D visual detection technology provides a feasible solution. In this paper, we propose a fundamental multi-source visual data detection method, as well as an accurate and robust fastener location and nut or bolt segmentation algorithm. By combining two-dimensional intensity information and three-dimensional depth information generated by the projection of line structural light, the locating of nut or bolt position and accurate perception of height information can be realized in the dynamic running environment of railway. The experimental results show that the static measurement accuracy in the vertical direction using the structural light vision sensor is 0.1 mm under the laboratory condition, and the dynamic measurement accuracy is 0.5 mm under the dynamic train running environment. We use dynamic template matching algorithm to locate fasteners from 2D intensity map, which achieves 99.4% accuracy, then use the watershed algorithm to segment the nut and bolt from the corresponding depth image of located fastener. Finally, the 3D shape of the nut and bolt is analyzed to determine whether the nut or bolt height meets the local statistical threshold requirements, so as to detect the hidden danger of railway transportation caused by too loose or too tight fasteners.
The detection of rail surface defects is vital for high-speed rail maintenance and management. The CNN-based computer vision approach has been proved to be a strong detection tool widely used in various industrial scenarios. However, the CNN-based detection models are diverse from each other in performance, and most of them require sufficient training samples to achieve high detection performance. Selecting an appropriate model and tuning it with insufficient annotated rail defect images is time-consuming and tedious. To overcome this challenge, motivated by ensemble learning that uses multiple learning algorithms to obtain better predictive performance, we develop an ensemble framework for industrialized rail defect detection. We apply multiple backbone networks individually to obtain features, and mix them in a binary format to obtain better and more diverse sub-networks. Image augmentation and feature augmentation operations are randomly applied to further make the model more diverse. A shared feature pyramid network is adopted to reduce model parameters as well as computation cost. Experimental results substantiate that the approach outperforms single detecting architecture in our specified rail defect task. On the collected dataset with 8 defect classes, our algorithm achieves 7.4% higher mAP.5 compared with YOLOv5 and 2.8% higher mAP.5 compared with Faster R-CNN.
Abstract. With the rapid development of computer vision, techniques of machine vision and visual inspection have been applied into the inspection of catenary on high-speed railways. Visual inspection systems have been developed and super-high-resolution images are captured to check the status of catenary components. Automatic recognition of defects becomes very important since the number of images is too huge to be manually checked one by one. However, it is not easy for the development of recognition algorithms on catenary components. There are many types of defects to be checked on different kinds of catenary components, but the number of defect images is too small in real world. In this paper, a solution was proposed and implemented. An on-site data acquisition system was designed and developed, and different types of defects were manually made on different catenary components beforehand. Finally, a visual inspection database was successfully constructed, including plenty of different kinds of catenary components, different types of defects, in different inspection conditions. The visual inspection database will be of great use in the development and test of recognition algorithms for catenary.
Image-based rail defect detection could be conceptually defined as an object detection task in computer vision. However, unlike academic object detection tasks, this practical industrial application suffers from two unique challenges, including object ambiguity and insufficient annotations. To overcome these challenges, we introduce the pixel-wise attention mechanism to fully exploit features of annotated defects, and develop a feature augmentation framework to tackle the defect detection problem. The pixel-wise attention is conducted through a learnable pixel-level similarity between input and support features to obtain augmented features. These augmented features contain co-existing information from input images and multi-class support defects. The final output features are augmented and refined by support features, thus endowing the model to distinguish between ambiguous defect patterns based on insufficient annotated samples. Experiments on the rail defect dataset demonstrate that feature augmentation can help balance the sensitivity and robustness of the model. On our collected dataset with eight defected classes, our algorithm achieves 11.32% higher mAP@.5 compared with original YOLOv5 and 4.27% higher mAP@.5 compared with Faster R-CNN.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.