A substation attribute segmentation method based on convolutional neural network
A convolutional neural network and substation technology, applied in the field of new neural networks, can solve problems such as difficulty in attribute segmentation tasks and failure to meet the requirements of image understanding, achieve high practicability and feasibility, prevent and eliminate power transmission faults, and ensure safety and smooth effect
Active Publication Date: 2018-11-02
STATE GRID JIBEI ELECTRIC POWER COMPANY +3
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] The images obtained in real life are often affected by changes in pose, scale, and illumination, which makes the attribute segmentation task very difficult. The main strategy of traditional attribute segmentation algorithms is to assign a single label to each pixel of the image. Conforms to the requirements of human for image understanding
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0022] The present invention will be further described below in conjunction with the accompanying drawings.
[0023] Such as figure 1 , figure 2 As shown, first, a substation image database containing seven types of images is established. The seven types of images are substations, transformers, switches, transmission lines, insulators, business halls, and people. All images are from the Internet and real monitoring equipment. The number of images in each category ranges from 60 to 140, and each category of images has great differences in background, angle, illumination, and scale. For each image, a binary attribute table is obtained through manual annotation, which indicates the attribute category contained in the image, for example, {1,0,0,1,0,0,1} indicates that the An image contains three types of attributes: substation, transmission line, and person; at the same time, for each image, each attribute is manually labeled to obtain a semantic segmentation, and each semantic...
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
Login to View More Abstract
The invention discloses a substation attribute segmentation method based on a convolutional neural network, which establishes a substation image database containing seven types of images, constructs an attribute table for the image and manually annotates semantic segmentation, pre-trains, learns attribute classification and attribute segmentation volume Convolutional neural network, and then attribute classification and attribute segmentation of images through convolutional neural network. A convolutional neural network-based substation attribute segmentation method provided by the present invention can effectively prevent and eliminate power transmission faults and ensure safe and smooth power supply; at the same time, it can effectively relieve the pressure of human monitoring and achieve intelligent monitoring in the true sense. On the establishment of the substation image database, a large number of tests have been carried out, and the results show that the new technology based on the deep convolutional neural network proposed by the present invention has very high practicability and feasibility.
Description
technical field [0001] The invention relates to a substation attribute segmentation method based on a convolutional neural network, and belongs to the technical field of novel neural networks for attribute segmentation. Background technique [0002] In recent years, the image classification task only gives a result label for the entire test image. For example, for a substation picture to be tested, the image classification task is to accurately output the label "substation" for the test picture, so that it can It is understood that this is an image of a substation; and attribute segmentation is different. Attribute segmentation needs to assign a label to each pixel of the image, and then gather pixels with the same label together, so that the image can be divided into multiple blocks. That is, assigning multiple attributes to the image, such as the image of the substation to be tested mentioned above, can be interpreted as: this is an image with multiple attributes such as s...
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
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/66
CPCG06V30/194
Inventor 吴佳苏丹郝小龙袁卫国彭启伟李环媛罗旺刘超余磊高崧冯敏
Owner STATE GRID JIBEI ELECTRIC POWER COMPANY