Product introduction
Alibaba Cloud Optical Character Recognition (OCR) provides an end-to-end technology system for text localization, recognition, and comprehension in images. The service effectively detects text in complex scenarios, such as skewed images, offset text, overlapping seals, and watermark interference. It supports various use cases, such as general text recognition for handwriting, E-commerce images, and tables. It also supports recognition for personal and enterprise credentials, education exam questions, and ticket and receipt verification. The service offers both public cloud API calls and private deployment modes to meet diverse business needs.
Product features
A single API integrates 59 recognition capabilities across six major scenarios: general text, personal credentials, vehicle and logistics documents, tickets and receipts, enterprise qualifications, and mixed invoices.
Detects text in general images, Chinese and English handwriting, and tables with or without lines. It also supports features such as low confidence level filtering and pattern content detection.
Personal credential recognition
Detects text from personal credentials such as ID cards, passports, household registration books, social security cards, and bank cards.
Enterprise qualification recognition
Detects text from qualification documents used in business operations, such as business licenses, trademark registration certificates, bank account opening certificates, and cosmetics business permits.
Ticket and receipt recognition
Detects text from various common tickets and receipts required for expense reimbursement, such as mixed invoices, value-added tax (VAT) invoices, general machine-printed invoices, train tickets, ride-hailing itineraries, and shopping receipts.
Detects text from vehicle and logistics-related documents, such as vehicle registration certificates, driver's licenses, license plates, vehicle identification numbers (VINs), and logistics waybills.
Education scenario recognition
Detects text from printed mathematical formulas and exam papers, and supports question segmentation and oral arithmetic evaluation.
Detects text in languages such as English, Japanese, Russian, Korean, and Thai.
Ticket and receipt verification
Verifies the authenticity of business licenses and invoices.
Product workflow
Step 1: Try the service without logging on
Go to the Alibaba Cloud OCR Experience Center > Experience Center. You can use the visual interface to upload a single image and try the OCR features. No account registration or logon is required. You can try all OCR services for free an unlimited number of times.
The top of the page provides multiple recognition type tabs (such as ID card recognition, bank card recognition, and household registration book recognition). Select the required recognition type, enter an image URL or click Upload Image at the bottom to add an image, and then click Detect to view structured recognition results on the right side.
Step 2: Activate the service and get a free quota
If the product meets your business needs, you can go to the console to activate the required services. The OCR product offers 10 types of services. Each service must be activated separately at no charge. Depending on the activated service, you will receive a monthly free quota of 200 calls or a one-time free quota of 50 calls.
Select the corresponding service type tab (such as Personal Credential Recognition), and click the Activate Service button at the top of the page to activate the sub-services under that category (such as ID card recognition and international passport recognition). Each sub-service has a default concurrency limit of 10 QPS.
Step 3: Test API calls online
Before you integrate the software development kit (SDK) into your business system, you can test API calls online using the OpenAPI Developer Portal. The portal provides sample code in multiple languages and dependency information. This helps you understand the logic for calling the code and simplifies future API use.
Set the service region to China (Hangzhou).
Enter the URL of the image to be recognized in the Url parameter.
Set Type (image type) to Advanced, and enable Boolean parameters such as OutputFigure and OutputQrcode as needed.
Click Initiate Call. After a successful call, the right side displays status code 200 and the recognition result in JSON format, including fields such as RequestId and Data.SubImages.
Step 4: Integrate and call the SDK
The SDK encapsulates all the OCR services and includes the AccessKey pair information for caller identity verification. This removes the need for developers to manually construct signatures, which simplifies the development process and improves efficiency.