Image Recognition Method of Brake Beam Pillar Broken Fault of Railway Freight Car
A railway freight car and image recognition technology, which is applied in image analysis, image enhancement, image data processing, etc., can solve the problems of low detection accuracy and poor stability, and achieve the effect of unified operation standards, high accuracy, and fast speed
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
specific Embodiment approach 1
[0029] Embodiment 1: The specific process of the image recognition method for the fractured fault of the brake beam strut of a railway freight car in this embodiment is as follows:
[0030] Step 1. Create a sample data set;
[0031] Step 2: Based on the sample data set, find the optimal weight coefficient to obtain a trained SSD deep learning network; the specific process is as follows:
[0032] Step 21: Initialize the weight coefficient in a random way;
[0033] Step 22: Perform multi-scale feature extraction on the sample data set to increase the receptive field; SSD is a multi-scale feature map detection network structure, which extracts feature maps of different scales for detection. feature maps) can be used to detect small objects, while small-scale feature maps (later feature maps) are used to detect large objects. The receptive fields included in the feature map selected by the model are: 38, 19, 10, 5, 3, 1. For each feature map, four offset positions and N categor...
specific Embodiment approach 2
[0040] Embodiment 2: The difference between this embodiment and Embodiment 1 is that a sample data set is established in the first step; the specific process is:
[0041] Build high-definition equipment around the truck tracks respectively. After the truck passes through the equipment, obtain a full range of high-definition images on both sides, bottom and upper part of the truck; the image is a clear grayscale image; because the parts of the truck may be affected by rain, mud, oil, and white paint , black paint, foreign objects, ice and snow, chalk and other natural or man-made conditions. Also, there may be differences in images taken at different sites. As a result, the brake beam strut images vary widely. Therefore, in the process of collecting the image data of the brake beam strut, it is necessary to ensure the diversity, and try to collect all the brake beam strut images under various conditions.
[0042] In the truck bogies of different models, the shape of the brake...
specific Embodiment approach 3
[0053] Embodiment 3: The difference between this embodiment and Embodiment 1 or 2 is that the default box (anchor box) is generated in the steps 2 and 3; the specific process is:
[0054] For each feature map, generate k default boxes (default boxes [default boxes]) according to different sizes (scales) and aspect ratios (ratios);
[0055] The formula for calculating the size of each default box is:
[0056] Among them, m is the number of feature maps, s min is the size of the bottommost feature map, s max Default box size for the topmost feature map;
[0057] The aspect ratio of each default box is calculated according to the scale value, and the scale value is {1, 2, 3, 1 / 2, 1 / 3};
[0058] For the default box with a scale of 1, add an additional scale of the default box;
[0059] In the formula, s' k For the increased ratio ({1,2,3,1 / 2,1 / 3} there are a total of 5 default boxes (one number represents one), this is the size of the increased default box, 5+1=6 default bo...
PUM
Login to view more Abstract
Description
Claims
Application Information
Login to view more - R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap