Product Announcements

[New Feature] General Ticket Extraction Now Available

General Ticket Extraction uses a combination of OCR and large language model (LLM) technologies to extract information from long-tail tickets and ID documents that are not supported by standard OCR models. No training or configuration is required. This service improves the efficiency of enterprise automation.

[Upgrade] OCR Unified Recognition Adds New Recognition Capabilities

OCR Unified Recognition integrates multiple recognition capabilities into a single API to meet the diverse needs of enterprises. This release adds 15 new capabilities across three scenarios. OCR Unified Recognition now supports 59 capabilities across six scenarios to deliver broader recognition coverage.

• Supports six major categories: general text recognition, personal ID recognition, vehicle and logistics recognition, invoice and receipt recognition, business qualification recognition, and mixed-document recognition.

• Mixed-document recognition supports all document types that are listed in the table below, except for vehicle VIN codes, license plates, electronic waybills, international passports, international IDs, official seals, medical device operation licenses, medical device manufacturing licenses, cosmetics manufacturing licenses, international business licenses, trademark registration certificates, food operation licenses, food manufacturing licenses, Class II medical device operation filing certificates, or bank account opening licenses.

• The following recognition capabilities are supported:

  Scenario	Recognition capability
  General text recognition (8 types)	General Text Recognition – High Accuracy Edition	General Text Recognition – Basic Edition	Handwriting
  	E-commerce image text	Multilingual text	Tables
  	QR codes	Barcodes
  Personal ID recognition (13 types)	ID cards	Bank cards	Social security cards
  	Household registration book – front page	Household registration book – permanent resident page	Birth certificates
  	Chinese passport	International passports	Permits for traveling to Hong Kong (China), Macao (China), and Taiwan (China)
  	Mainland Travel Permits for Hong Kong and Macao Residents	Hong Kong (China) ID cards	International ID cards
  	Real estate ownership certificates
  Vehicle and logistics recognition (7 types)	Vehicle registration certificates	Driver’s License	License plates
  	Vehicle VIN codes	Vehicle conformity certificates	Motor vehicle registration certificates
  	Electronic waybills
  Invoice and receipt recognition (19 types)	VAT invoices	Unified motor vehicle sales invoices	Fixed-amount invoices
  	Roll-type VAT invoices	General computer-printed invoices	Non-tax revenue invoices
  	Unified second-hand vehicle sales invoices	Taxi receipts	Tax payment certificates
  	Bank acceptance bills	Passenger transport tickets	Train tickets
  	Air travel itineraries	Hotel transaction records	Shopping receipts
  	E-commerce order pages	Ride-hailing trip itineraries	Payment detail pages
  	Toll road and bridge fee invoices
  Business qualification recognition (11 types)	Business License	Official seals	Medical Device Operating License Identification
  	Medical device manufacturing licenses	Cosmetic Production License Detection	International business licenses
  	Trademark Registration Certificate Recognition	Food Business License Detection	Detecting Food Production Licenses
  	Class II medical device operation filing certificates	Bank Account License Detection
  Mixed-document recognition (1 type)	Mixed ticket and receipt documents

[New Feature] OCR Unified Recognition Now Available

OCR Unified Recognition is a new service from the Alibaba Cloud OCR team. A single API integrates 44 recognition capabilities, including both single-scenario and mixed-document recognition, to meet diverse requirements. This improves the ease of integration, usability, and efficiency, and lowers the entry barrier for customers who require multiple OCR capabilities.

• It covers six major categories: general text recognition, personal ID recognition, vehicle and logistics recognition, invoice and receipt recognition, business qualification recognition, and mixed-document recognition.

• Mixed-document recognition supports all document types that are listed in the table below, except for vehicle VIN codes, license plates, international passports, international ID cards, or official seals.

• The following recognition capabilities are supported:

OCR Document Self-Learning Fully Commercialized

The OCR Document Self-Learning platform will be fully commercialized on August 23, 2023 at 00:00 (UTC+8). Charges will apply to the following six services: document and receipt information extraction, table information extraction, long-document information extraction, custom key-value templates, custom table templates, and prebuilt capabilities. After commercialization, the document and receipt information extraction, table information extraction, and long-document information extraction services will be billed on a pay-as-you-go basis according to the training and deployment time and the number of API calls. The custom key-value and custom table template services support both pay-as-you-go and resource plan billing methods. The prebuilt capabilities service supports only the pay-as-you-go billing method.

After you activate the service, the pay-as-you-go billing method is used by default. Each service includes a specific free quota for testing. If you have purchased a resource plan that is still within its validity period, it can be used to offset the charges for the corresponding service types.

[Upgrade] OCR Adds Pay-as-you-go Billing