Stable and reliable high-precision OCR service
Based on Tencent's deep learning technology and massive amounts of data, Tencent Cloud TI X-Vision: Optical Character Recognition (TIXOCR) provides printed and handwritten character recognition services for a wide variety of cards, certificates and documents and supports custom templates, greatly improving data entry efficiency and reducing the cost of use.
Leveraging Tencent's powerful deep learning algorithms and leading OCR technology, TIXOCR achieves an average recognition accuracy rate of over 95% for printed text and over 90% for handwritten text and has an average recognition speed of 2 ms per character for GPU and 6 ms per character for CPU. In addition, its algorithm models are continuously optimized and iterated.
With its high availability and performance proven effective by massive volumes of end users and complicated application scenarios, TIXOCR has been widely adopted by Tencent's core businesses such as WeChat, QQ, Tencent Social Ads and WeBank as well as prestigious customers like Kuaishou, SF Express, SinoTrans and Dianping.com.
With a simple and intuitive demo and high compatibility, TIXOCR not only has APIs that can be called directly, but also provides a variety of SDKs for developers.
TIXOCR can recognize printed and handwritten text on dozens of types of cards and documents such as ID cards, road-worthiness certificates, driver's licenses, business licenses, business cards and invoices and offers custom templates for text recognition to meet diverse requirements in different scenarios.
TIXOCR can recognize Simplified and Traditional Chinese characters, letters, digits and special characters on printed images with any layout and photos in various scenarios and return their corresponding positions. In addition to Chinese and English, it now supports 18 other languages including Japanese, Korean, Spanish, French, German and Portuguese.
TIXOCR boasts enhanced recognition capabilities for handwritten text with "irregular, sloppy and fuzzy legibility." It can recognize handwritten Chinese characters, letters, digits and common characters in images with any layout in different scenarios.
TIXOCR supports the recognition of all fields on the front and back of a second-gen Chinese ID card, including name, gender, ethnicity, date of birth, address, citizen ID number, issuing authority and validity period, and features alarming for photocopied and temporary ID cards.
TIXOCR supports the recognition of all fields on a business license image, including registration number, legal representative, company name, address, business period, business scope and subject type.
TIXOCR supports automatic location and recognition of all fields on the pages of a road-worthiness certificate, including license plate number, vehicle type, owner, address, nature of use, brand and model, vehicle identification number, engine number, registration date, date of issuance and issuing authority.
TIXOCR supports automatic location and recognition of all fields on the primary page of a driver's license, including license number, name, gender, nationality, address, date of birth, date of initial issuance, permitted vehicle types and validity period.
TIXOCR supports automatic location and recognition of various fields on a business card, including name, phone number, mobile number, email address, company, department, title, website address, business address and QQ, WeChat and MSN account information.
TIXOCR can automatically recognize the card number and expiration date of mainstream bank cards and return information such as the issuing bank, card type and cardholder's name.
TIXOCR can automatically locate and recognize the fields on a vehicle license plate issued in Mainland China and return the geographical number and license plate number.
TIOCR can automatically recognize and collect the 17-character VIN with a recognition accuracy of over 99%, mainly applicable for industries such as auto insurance, used car trading and vehicle maintenance.
TIXOCR can automatically recognize and score rapid calculation questions, currently covering 14 types of questions within the K12 scope with a recognition accuracy of over 91%.
TIXOCR supports the recognition of all fields on a VAT invoice, including price after tax, invoice code, invoice number, total tax amount, total amount, password area, invoicing date, tax rate, purchaser's identification number and seller's identification number.
TIXOCR supports the recognition of key fields on a vehicle invoice, including machine number, machine code, invoice number, invoice code, ID number, invoicing date, vehicle identification number, seller's name, purchaser's name, price before tax, tax rate, price after tax (in words), price after tax (in numbers) and VAT amount.
TIXOCR can recognize the profile page of a passport issued in mainland China and output structured key fields, including birthplace, date of birth, nationality (in English), gender (in English), passport number, expiration date, issuing country, issuing place (in English), date of issuance, type, given name (in English) and surname (in English).
Identity Recognition in Finance
ID card recognition can be widely used in industries such as banking, insurance and securities where end user's identity requires verification to help reduce the workload of user data entry, increase efficiency and improve user experience.
Identity Recognition in Internet-based Businesses
In live video broadcasting and online car hailing scenarios, TIXOCR can verify the identities of VJ hosts and passengers to reduce the risk of violations and crimes.
Ecommerce Ad Image Recognition
In the ecommerce industry, general OCR can intelligently identify text information in ad images to filter illegal words and reduce the risk of violations.
Video Subtitles Recognition
General OCR can quickly identify subtitles in various videos to effectively distinguish whether the text content is compliant.
Traditional News Media
General OCR can help convert a large number of paper documents into editable format, which greatly facilitates text formatting and information retrieval.
Images in Publications
General OCR can help convert images in publications into editable format, which notably reduces the labor cost of publication classification and retrieval.
The express delivery industry generates high volumes of handwritten or printed waybills, and the recording and archiving of such information consumes a lot of manpower and material resources. General OCR can help effectively reduce labor costs and optimize business processes for improved efficiency.