Unlock instant, AI-driven research and patent intelligence for your innovation.

Rail surface defect detection method based on depth learning

A defect detection and deep learning technology, applied in the field of rail surface defect detection based on deep learning, can solve the problem of consuming a lot of manpower and material resources

Inactive Publication Date: 2018-12-21
CHANGSHA UNIVERSITY OF SCIENCE AND TECHNOLOGY
View PDF6 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, these methods have disadvantages. The detection work consumes a lot of manpower and material resources, and the final detection results need to be processed manually to make judgments.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Rail surface defect detection method based on depth learning
  • Rail surface defect detection method based on depth learning
  • Rail surface defect detection method based on depth learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0122] specific implementation plan

[0123] Step 1, make a dataset:

[0124] First of all, network training is required: the rail surface images collected on site need to be made into required data sets for unified processing, including training data sets and label data sets. The training data set is the rail body, and other background areas such as railway track sleepers and stones need to be removed from the acquired scene images, and only the rail body is detected. Considering the data training time comprehensively, it is most appropriate to set the size of the rail body image to 1250×55 (length×width, unit: pixel).

[0125] The production of the label data set established the standard for defining rail defects. Using image editing software, the defective area was designated as a yellow area (RGB: 255, 255, 0), and the remaining non-defective areas were designated as a black area (RGB: 0, 0 ,0), define the RGB value in the label data set used in the training of the convo...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a rail surface defect detection method based on depth learning. The invention overcomes the traditional non-intelligent rail surface defect detection method, combines the research results of depth learning, and proposes the rail surface defect detection based on semantic segmentation, which can be used for non-destructive and non-contact detection. The surface images of therails are made into the rails data set and sent to the designed neural network, The self-defined network based on semantic segmentation is used to train and learn, and finally the trained network isobtained, which is used to detect and mark the defect area of rail surface defect image. Combined with the image processing technology at the back end, the defect contour can be obtained, which can beused for intelligent recognition, achieve high-precision detection and reduce the purpose of manual intervention.

Description

technical field [0001] The invention utilizes the field of image processing, and in particular relates to a method for detecting defects on the surface of rails based on deep learning. Background technique [0002] At present, the commonly used methods for the detection of rail surface defects in my country include manual detection, eddy current coil detection and ultrasonic detection. However, these methods have disadvantages. The detection work consumes a lot of manpower and material resources, and the final detection results need to be processed manually to make judgments. Considering that ultrasonic testing and eddy current coil testing will contact with rail surface defects, physical and chemical changes may occur, which further expands the defect area. Therefore need to improve it. [0003] Glossary: [0004] 1. The RGB color mode is a color standard in the industry. It is obtained by changing the three color channels of red (R), green (G) and blue (B) and superimpo...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06T7/00G06T7/12G06T7/181G01N21/88
CPCG01N21/8851G01N2021/8887G06T7/0004G06T7/12G06T7/181
Inventor 张辉宋雅男刘理钟杭梁志聪
Owner CHANGSHA UNIVERSITY OF SCIENCE AND TECHNOLOGY