Document snapshot method and device and computer storage medium
What is Al technical title?
Al technical title is built by PatSnap Al team. It summarizes the technical point description of the patent document.
A document and image technology, applied in the field of document capture, can solve problems such as difficulty in document entry
Pending Publication Date: 2020-12-04
CHINA PING AN PROPERTY INSURANCE CO LTD
View PDF0 Cites 0 Cited by
Summary
Abstract
Description
Claims
Application Information
AI Technical Summary
This helps you quickly interpret patents by identifying the three key elements:
Problems solved by technology
Method used
Benefits of technology
Problems solved by technology
[0003] In view of this, a document capture method is provided to solve the problem of difficult document entry
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
Click on the blue label to locate the original text in one second.
Reading with bidirectional positioning of images and text.
Smart Image
Examples
Experimental program
Comparison scheme
Effect test
no. 1 example
[0042] refer to figure 1 , figure 1 It is the first embodiment of the document capture method of the present application, and the document capture method includes the following steps:
[0043] Step S110: extracting the first document image set in the video stream through the document classification model;
[0044] The classification model is a model for classification obtained by using a training set for training using a machine learning method. Through the training of a large amount of data, it can classify all the data by itself, and the objects under each classification will have similar characteristics and corresponding characteristic parameters. For this application, the training set is pictures of documents in different situations, such as situations where the image is too close, too far, overexposed, very dark, or has a complex background in the collection field of view. For the documents required by the application scenario, the training set should include them to f...
no. 2 example
[0095] refer to figure 2 , figure 2 It is the second embodiment of the document capture method of the present application, and the document capture method includes the following steps:
[0096] Step S210: intercepting the video stream during the video call.
[0097] The video call process under the application scenario of intercepting a vehicle accident can be carried out on the claim adjuster's mobile phone, tablet, work computer platform, or any terminal with the function of intercepting and recording video. The intercepted video stream is the basis for later classification and recognition.
[0098] Step S220: Collect video frames from the video stream at a preset period or a preset frame number interval.
[0099] It can be understood that, in order to reduce resource consumption, it is not necessary to continuously acquire each frame of images for classification processing. In one embodiment, at the very beginning, for example, when a video call is established, an ima...
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 document snapshot method and device and a computer storage medium. The method comprises the steps of extracting a first document image set in a video stream through a document classification model; identifying each image in the first document image set through a document identification model to obtain a second document image set meeting the image transformation requirement, wherein the document identification model is a convolutional neural network model adopting artificial intelligence, and a full connection layer of the document identification model comprises a position information parameter and a position information confidence coefficient parameter corresponding to a document area in a first document image; and performing image transformation on each image inthe second document image set to obtain a third document image set meeting information extraction requirements. According to the method, the problem of difficulty in document entry is solved, and theoperation experience of user document entry is improved.
Description
technical field [0001] The invention relates to the technical field of intelligent capture, in particular to a document capture method. Background technique [0002] The current document text recognition OCR (Optical Character Recognition, Optical Character Recognition) function is processed by users taking photos of documents and uploading them to the cloud server, but the documents uploaded by users have problems that do not meet the shooting requirements, such as: incomplete photography, Shooting at a long distance, the image is seriously reflective, or the shooting environment is too dark, etc. The above problems often lead to text recognition failures, or document entry errors. For this kind of problem, the usual practice in the industry is to infer the reliability of the document recognition result based on the confidence threshold of text recognition, and refuse to return the result for the document with too low confidence in the recognition result. However, it take...
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
Application Date:The date an application was filed.
Publication Date:The date a patent or application was officially published.
First Publication Date:The earliest publication date of a patent with the same application number.
Issue Date:Publication date of the patent grant document.
PCT Entry Date:The Entry date of PCT National Phase.
Estimated Expiry Date:The statutory expiry date of a patent right according to the Patent Law, and it is the longest term of protection that the patent right can achieve without the termination of the patent right due to other reasons(Term extension factor has been taken into account ).
Invalid Date:Actual expiry date is based on effective date or publication date of legal transaction data of invalid patent.
Login to view more 
Login to view more