Many companies manage various image formats and perform model retraining to improve document recognition rate, ultimately reducing manual labor and ensuring process automation and speed. TwinReader provides process automation to extract, understand, and formalize data through its own hybrid algorithm based on accumulated experience.
Unstructured image
and/or e-document
Text recognition
Categorization > Understanding > Analyze
Converts into standardized data
When an image is put in User’s UI that reflects TwinReader’s feature, TwinReader uses its API to process the image, extract information and validate accuracy to convert the image into data that the user can utilize.
Service
Management
Hardware
monitoring
API
Custom Models
Field Schema (Define extraction targets)
User Defined Plugin
Processes images within tables
Summarizes text and corrects recognition results
Strives for 100% document automation
For accurate recognition and understanding
Text location searching and recognition Table recognition Text property recognitionFor user convenience
Extraction Engine & Multimodal Model Classification of document type Provide a recognition model by document types Workbench Administrator supportFor high performance
Verification of recognition results Correction of recognition results Works in combination with different systemsFinance
Product contracts/ Service application forms / Financial forms / Customer care documentation / Reports
Medical
Medical expenses receipt / Personal health documents
Service
Employment or rental contracts /Terms and conditions and legal provisions / User’s manuals
Sales/Finance
Transactions contract management / Financial statements / Invoices / Receipts and payment management / Purchase request / Cost estimates and contract management / Company management documents / Asset management documents
Human
Resources
Report management / Attendance Management / Work benefits/welfare management /Employee information management / Information recognition and masking
Manufacturing/
Logistics
Inspection of manufacturing labeling / Equipment manual / Demand and supply planning / Production information management / Price and inventory management / Job scheduling / Process scheduling / Processing customs documents
Recognition and automation of insurance claims for ‘A’ insurance company
Provided automated processing of all insurance claims – up to 7.5million cases a year
Insurance claim recognition accuracy
Total costs saved over 5 years after implementation
5billion KRW(approx. 3.5mil USD)Computerizing the corporate evaluation of data of ‘B’ credit rating agency
Standardized the information contained in different forms sent by enterprise customers, and computerized each them
Average processing speed per image
2.6secAverage recognition accuracy
97.2% and overRecognition and automation of insurance claims for ‘C’ insurance company
No. of processed of images per hour
More than 4,000Document categorization success rate
95% and over2-3F, Infostorm Building, Seolleung-ro 525 Gangnam-gu, Seoul | TEL(02-558-8300) | contact@agilesoda.ai
Copyright © AgileSoDA
.