Function Introduction
-
Model management
Supports displaying a list of all models, provides a new model function, can import and export models, and each model can display template ID, template name, training time and other model details
-
Model training
Automatically/manually select text and fixed-position content in the picture, automatically analyze the template image, quickly set the reference area and identification area, and edit, test, publish, delete and other operations for each model in just a few minutes Completed, the template can be called after it is officially released.
-
Model annotation
Automatically perform full-page analysis and recognition. During training, the reference area and recognition area can be marked based on the results, and the template image can be rotated, enlarged, reduced, moved, etc.
-
Automatic analysis of details list without frame line/with frame line
Automatic column analysis can be performed on detail areas of unframed/framed lists, and support for adding, deleting, and modifying automatically analyzed detail columns with high recognition rate.
-
The platform has built-in rich recognition cores
Supports recognition of simplified Chinese, handwriting, traditional Chinese, printed mixed samples, English and numbers, with high recognition rate, and users can set the recognition fields by themselves.
-
Custom classifier
For documents in unique corporate formats that need to be classified, a classifier can be created by uploading 30 training images for each category to achieve automatic classification of images in different formats.
-
Output structured data
Returns JSON and XML structured data of various forms/cards to facilitate editing and achieve instant synchronization of electronic and paper document information
-
Advanced settings
More advanced settings can be made for each recognition area, including line limit, character set limit, and regular expression limit to improve the recognition rate.
-
Intelligent extraction of key field information
Supports extracting fixed rows or fixing the same field information from multiple rows of content
-
Character replacement
Built-in regular expressions to proofread and replace error-prone text
-
Complex irregular sample model training
Supports recognition training of complex irregular sample models (such as medicine bottle instructions)
Product Superiority
Smart color filter
- Can intelligently filter red and blue colors in samples to improve recognition rate
A variety of mature models are embedded and can be called directly
- Embedded with mature models such as traditional cards and bills, you can directly select and use them with high recognition rate
Flexible editing and modification
- If you are satisfied with the results of multiple tests, you can publish it. If you are not satisfied with the results, you can go back and continue editing.
Deep learning algorithm technology
- Complex samples such as tilt, rotation, photo distortion, background, and incomplete documents can all be identified.
Supports recognition of images in multiple file formats
- Supports uploading of various file formats such as JPEG, PNG, PDF, etc., and supports form recognition in various acquisition forms such as taking photos, scanning, printing, and online banking.
Private deployment
- Supports privatized deployment and is deployed to the user's local server to ensure data privacy
Application Scenarios
- Electronic cards and documents
- Medical bill identification
- Business analysis
- Bank note identification
-
Electronic cards and documents
Apply the OCR training platform to carry out structured identification of various cards and documents such as ballot cards and admission tickets with non-uniform formats to achieve electronic management of various cards and documents. It can be applied to paper document classification and archiving and information statistical analysis. , key content extraction and other scenarios, effectively reducing manual entry costs and greatly improving information management efficiency.
-
Medical bill identification
Use the OCR training platform to create templates for commonly used medical bills such as inspection reports, outpatient billing, and hospitalization billing to achieve automatic classification and structured identification. It can be applied to scenarios such as medical data analysis, medical expense reimbursement, and electronic medical bills. Realize the automatic identification and entry of relevant bill information, effectively reduce labor costs, and greatly improve the level of medical informatization.
-
Business analysis
Bank statement is an important basis for understanding the operation of the enterprise. Through the OCR training platform, it can realize the intelligent extraction of bank statement information in various formats, assist users to review whether the account statement period is complete, and facilitate analysis and verification of whether the company has fraudulent statements and fictitious profits. etc., to solve problems such as numerous running accounts, difficulty in unifying the layout, and difficulty in obtaining data.
-
Bank note identification
In the process of bank account opening, credit approval, post-supervision and other business approval processes, staff need to review a variety of materials. The time span is long, it is difficult to share electronic and paper data, and manual review and approval efficiency is low. Use the OCR training platform to create templates for common bank bills such as business vouchers, applications, receipts, statements, bills of exchange, bank statements, etc., to achieve automatic classification and structured identification, which can be applied to bank account opening, credit, post-supervision and other scenarios , realize automated information extraction and intelligently enter it into the business system, effectively reducing labor costs and controlling business risks.
Case
-
- How does Traditional Hotels Transform into Smart Hotels
-
- The Combination and Verification of Person and Document in the Hotel
-
- ID Card Reader Increases Hotel Occupancy
