Text detection method based on central point supervision information
A text detection and center point technology, applied to instruments, biological neural network models, calculations, etc., can solve problems such as increased detection difficulty and unclear images, and achieve the effect of reducing training difficulty, solving content detection problems, and avoiding adhesion
Inactive Publication Date: 2021-07-02
STATE GRID HEBEI ELECTRIC POWER CO LTD +1
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] The present invention provides a text detection method based on central point supervision information, which is used to solve the problem in the prior art that the collected images are unclear due to improper printing and / or storage of train tickets and increase the difficulty of detection
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
[0039] In order to make the object, technical solution and advantages of the present invention more clear, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, and are not intended to limit the present invention.
[0040] Such as Figure 1-2 As shown, the main modules of the technical solution of the present invention include as follows: the system includes three parts, the first part is a picture feature extraction module based on ResNet50; the second part adopts FPN to carry out information fusion of different scales; the third part is based on fusion As a result, five types of prediction results (text area, text centerline area, disk radius, and disk sin value) are generated, and the striding method proposed by this method is used for post-processing operations. In the first part, we use...
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 text detection method based on central point supervision information, and relates to the technical field of bill text detection, in particular to the text detection method based on the central point supervision information, which comprises the following steps: searching a boundary of a train ticket; performing feature extraction on the train ticket picture through a ResNet50 convolutional neural network; fusing the extracted feature maps through an FPN network; and generating five different outputs through the FPN: a text area, a text center line area, a disc radius, a sin value of the disc and a cos value of the disc to generate a final detection result. The invention is mainly used for detecting the rule text in the train ticket. The text is divided into sub-regions (discs) for prediction, so that the training difficulty is reduced, and the FPN module enables the network to have more flooding performance. And the adhesion problem caused by directly utilizing a text region is avoided by utilizing a disk representation method. The method can effectively solve the problem of content detection in the train ticket.
Description
technical field [0001] The invention belongs to the technical field of bill text detection, in particular to a text detection method based on central point supervision information. Background technique [0002] Train tickets are an important part of corporate travel reimbursement, which contains information necessary for reimbursement such as the starting station, travel date, and amount. However, at present, applicants mainly fill in and enter information manually, which is very time-consuming and labor-intensive. , if the deep learning technology can be applied to the reimbursement of train tickets to realize the automatic extraction of information, it will greatly save the cost of human resources. [0003] The automatic extraction process of train ticket information includes two stages of detection and recognition. As a basic step in the whole process, the detection link faces more problems. The current ticket is mainly printed in ink, and there will be problems such as...
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 Applications(China)
IPC IPC(8): G06K9/00G06K9/32G06K9/62G06N3/04
CPCG06V30/412G06V20/62G06N3/045G06F18/253
Inventor 刘义江陈蕾侯栋梁池建昆范辉阎鹏飞魏明磊李云超姜琳琳辛锐陈曦杨青沈静文吴彦巧姜敬檀小亚师孜晗
Owner STATE GRID HEBEI ELECTRIC POWER CO LTD