Unified OCR

Updated at:

Introduction

Unified OCR is a new product from Alibaba Cloud's Optical Character Recognition (OCR) team. A single API integrates 59 recognition capabilities for various scenarios, making integration simpler and more efficient. This lowers the barrier to entry for customers who need to integrate multiple OCR capabilities.

Features

  • Recognize documents from 59 single-document and mixed-document scenarios with a single API.

  • Covers six major categories: general text, personal IDs and documents, vehicle and logistics documents, invoices and receipts, business qualifications, and mixed documents.

  • Mixed document recognition supports combinations of images from the personal ID, vehicle and logistics, invoice and receipt, and business qualification categories. These categories are detailed in the table below. However, the following document types are not supported for mixed recognition: Vehicle VIN, License Plate, E-waybill, International Passport, International ID Card, Official Seal, Medical Device Business Permit, Medical Device Manufacturing Permit, Cosmetics Manufacturing Permit, International Business License, Trademark Registration Certificate, Food Business Permit, Food Production Permit, Class-II Medical Device Operation Filing Certificate, or Bank Account Opening Permit.

The following figure shows the supported detection capabilities:

Scenario

Capability Type

General Text Recognition

(8 types)

General Text Recognition (High-accuracy Edition)

General Text Recognition (Basic Edition)

Handwritten Text

E-commerce Image Text

Multi-language Text

Table

QR Code

Barcode

Personal ID and Document Recognition

(13 types)

ID Card

Bank Card

Social Security Card

Household Register (Front Page)

Household Register (Resident Page)

Birth Certificate

Chinese Passport

International Passport

Exit-Entry Permit for Travelling to and from Hong Kong, Macao and Taiwan

Mainland Travel Permit

Hong Kong (China) ID Card

International ID Card

Real Estate Title Certificate

Vehicle Identification for Logistics

(7 types)

Vehicle License

Driver's License

License Plate

Vehicle VIN

Vehicle Inspection Certificate

Motor Vehicle Registration Certificate

E-waybill

Invoice and Receipt Recognition

(19 types)

VAT Invoice

Unified Invoice for Motor Vehicle Sales

Quota Invoice

Rolled VAT Invoice

General Machine-printed Invoice

Non-tax Revenue Invoice

Unified Invoice for Used Car Sales

Taxi Receipt

Tax Payment Certificate

Bank Acceptance Bill

Passenger Ticket (Bus/Ferry)

Train Ticket

Air Travel Itinerary

Hotel Folio

Shopping Receipt

E-commerce Order Page

Ride-hailing Itinerary

Payment Details Page

Toll Invoice

Business Qualification Recognition

(11 types)

Business License

Official Seal

Medical Device Business Permit Recognition

Medical Device Manufacturing Permit Recognition

Cosmetics Manufacturing Permit Recognition

International Business License

Trademark Registration Certificate Recognition

Food Business Permit Recognition

Food Production Permit Recognition

Class-II Medical Device Operation Filing Certificate Recognition

Bank Account Opening Permit Recognition

Mixed Document Recognition

(1 type)

Mixed Documents

Note:

  • The "General Text Recognition (High-accuracy Edition)" capability is the same as the existing "Full Text Recognition (High-accuracy Edition)" but offers more configuration items than the original API.

  • The "General Text Recognition (Basic Edition)" capability is the same as the existing "General Text Recognition" but offers more configuration items than the original API.

Benefits

  • Compatible with multiple data types

    You can use a single API to recognize various data types across multiple scenarios. This eliminates the need to integrate multiple APIs and reduces integration costs.

  • Clear billing

    Unified OCR supports pay-as-you-go billing and shared resource plans. Multiple capabilities can consume from the same resource plan, which simplifies purchasing.

  • High real-time performance

    The service runs on Alibaba Cloud's self-built Elastic Algorithm Service (EAS) online clusters. Continuous optimization of inference technology provides a low-latency service with elastic scaling.

  • Stable service

    The service scales elastically based on call volume to ensure high extensibility. Continuous algorithm iterations and optimizations do not affect service stability for customers.

Scenarios

  • Image content moderation

    You can use general-purpose APIs to recognize content in various scenarios for content moderation. This lets you detect non-compliant behavior promptly and significantly reduce manual labor costs. This service is widely used in e-commerce content administration scenarios.

  • Logistics, transportation, and automotive

    This service is useful in scenarios such as used car transactions, smart parking, logistics, courier stations, and car insurance. The service provides recognition and verification for personal identities, vehicle information, and courier details. It supports text recognition for vehicle licenses, driver's licenses, license plates, VINs, e-waybills, ID cards, and vehicle inspection certificates.

  • Finance, banking, and insurance

    You can use artificial intelligence recognition services for scenarios such as remote account opening, identity verification, information entry, contract and policy digitization, and bank statement or financial report data entry. This greatly reduces manual labor costs. The service supports text recognition for ID cards, passports, travel permits, bank cards, official seals, and tables.

  • Government and healthcare

    This service is useful in scenarios such as financial reimbursement, paper-to-digital conversion of documents, volumes, and contracts, medical billing, pharmacy management, and resume or paper scanning and editing. The service converts non-text formats, such as images and scans, into editable text. It supports text recognition for VAT invoices, invoice verification, various receipts and vouchers, full text (high-accuracy), handwriting, and tables.

  • Retail, internet, and e-commerce

    This service provides text recognition from images for scenarios such as content administration of posters, flyers, and product pages, qualification verification for merchant onboarding, and business opportunity or brand mining. The service supports recognition capabilities for e-commerce images, business licenses, real estate certificates, bank account opening permits, tables, high-accuracy text, and general text.

Contact us

If you have any questions or feedback, or if you want to communicate directly with our engineers, contact us.

You can reach us in the following ways:

  • Join our official DingTalk group: 35208328

  • OCR Online Support

  • Email our business contact at ocr_support@list.alibaba-inc.com and include your contact information.