Text recognition method and system based on comparative learning
A text recognition and text technology, applied in the field of text recognition, can solve problems such as poor performance of text recognition methods
Active Publication Date: 2022-01-11
XIAMEN MEIYA PICO INFORMATION
View PDF3 Cites 4 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, current contrastive learning methods are only suitable for image classification, image segmentation, and image recognition that use the entire image as a single instance, and do not perform well in text recognition methods that contain sequential
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
[0036] The application 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 related inventions, rather than to limit the invention. It should also be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings.
[0037] It should be noted that, in the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and embodiments.
[0038] figure 1 An exemplary system architecture 100 of a method for text recognition based on contrastive learning to which the embodiment of the present application can be applied is shown.
[0039] Such as figure 1As shown, the system architecture 100 may include termin...
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 provides a text recognition method and system based on comparative learning, and the method comprises the steps: carrying out the data enhancement of each sample, inputting the data into a convolutional network, carrying out the recognition training, generating a recognition model, building a basic encoder based on the recognition model, and calculating and outputting a feature sequence; inputting the feature sequence into an instance mapping function to generate a corresponding instance, mapping the instance into a plurality of sub-instances, taking all the sub-instances as sub-elements in a comparison loss function to carry out comparison learning, and feeding back a result to the convolutional network for updating the convolutional network; and obtaining a tagged text image sample containing text information, inputting the tagged text image sample into the basic encoder, and adjusting parameters of the convolutional network until the recognition model converges. According to the method, comparative learning is applied to each element of the sequence, effective representation information is learned by fully utilizing unlabeled data, modeling is carried out based on a self-supervised comparative learning method, and the recognition effect is remarkably improved.
Description
technical field [0001] The invention relates to the technical field of text recognition, in particular to a text recognition method and system based on contrastive learning. Background technique [0002] In recent years, deep learning has been widely used in the fields of image classification, object detection and segmentation, etc., and it has also brought about problems such as high cost of data labeling and long time consumption. How to make good use of a large amount of unlabeled data for self-supervised learning has become a problem. current research hotspot. Self-supervised learning can mine its own supervision information from large-scale unlabeled data sets, and train the network through this structured supervision information, so that it can learn valuable representation information for downstream tasks. [0003] Remarkable results have been achieved in semi-supervised computer vision applications such as image classification, object detection, and segmentation, wh...
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): G06V10/22G06V30/10G06V10/74G06K9/62G06N3/04G06N3/08
CPCG06N3/088G06N3/045G06F18/22
Inventor 刘彩玲吴婷婷赵建强高志鹏汪泰伸陈德意
Owner XIAMEN MEIYA PICO INFORMATION