Document Reader SDK 8.3 — September 25, 2025
IMPORTANT changes
- Introduced age and gender estimation. This feature is available only when Integration with Face API is enabled. By default, age and gender estimation is disabled. The feature is now in beta state and will be fully rolled out in the next releases. The feature analyzes a person’s face in a document photo to predict its approximate age and gender, then automatically compares these predictions to the document’s data. This adds an extra layer of verification by ensuring consistency between biometric evidence and official document details, helping to detect mismatches and reduce the risk of fraud.
- Added the light version of Web Component –
"BoundsLite"
mode. This option provides only the functionality of locating document boundaries with simplified algorithms that will be useful for low-end devices. - Significantly improved performance and accuracy of document processing when working with Regula mobile devices, such as 1120 and 7310, in the
FullAuth
scenario. Improvements include better exposure adjustment depending on the environment condition and documents’ database description, UV images normalization for high-quality results, automatic adaptation to any IR camera’s position and field of view, and environment analysis with corresponding messages (“too dark” or “too bright”). - Improved RFID chip data reprocessing to enhance security and make it less dependent on the client side, ensuring reprocessing regardless of the version in which the data was collected. Therefore, it is recommended to update to this version.
Core SDK
Document Type Recognition
- Fixed issue with document type probability calculation, caused by introducing an additional step to resolve ambiguities between different series of the same document that differ only in IR/UV security features, based on the date of issue. Example: latest Germany ePassports.
OCR
- Fixed issue where unnecessary spaces were added between characters, resulting in outputs like
“T E S T”
instead of“TEST”
.
MRZ
- Updated parsers:
- Bonaire, Sint Eustatius, and Saba ID Card
- Ecuador ID Card
- France ID Card
- Moldova ID Card
- Mozambique ID Card
- Saint Barthelemy DL
- United States ID Card
Barcode
- Added new parsers:
- Columbia ID Card
- Mexico DL
- Philippines Police Card
- Updated parsers:
- Columbia ID Card
- Honduras DL
- Libya Visa
- Philippines Digital ID Card
- United States ID Card and DL
Text data parsing and validation
- Added an exception for French ID cards originally issued with a 10-year validity between 2004 and 2014, later extended by 5 years.
- Fixed issue with the document expiration validity status for Bulgarian documents issued with an indefinite term.
- Fixed issue with splitting surnames and given names into separate text fields.
- Fixed issue with removing gender prefixes from given names.
- Fixed issue with initiating the document maximum validity period analysis that verifies the date of issue and expiry date comply with the maximum allowed validity period.
- Fixed issue with gender value extraction from barcodes for USA and Canadian documents to match the visual zone value that are encoded according to the AAMVA standard, where the value is
N
instead ofX
. - When a date is not fully presented in the MRZ (for example, if the date of issue is missing a day) but the complete value is available in the Visual Zone, the value from the Visual Zone will be used for the MRZ value instead of adding
“01”
as the default day specified by ICAO.
Authenticity
- Introduced age and gender estimation. This feature is available only when Integration with Face API is enabled. By default, age and gender estimation is disabled. The feature is now in beta state and will be fully rolled out in the next releases. The feature analyzes a person’s face in a document photo to predict its approximate age and gender, then automatically compares these predictions to the document’s data. This adds an extra layer of verification by ensuring consistency between biometric evidence and official document details, helping to detect mismatches and reduce the risk of fraud.
- Added new property
processParam.strictSecurityChecks
, when enabled, marks security checks that don’t meet minimum requirements as"Failed"
(instead of"WasNotDone"
), which causes the overall security status to beFailed
. - Added the return of a negative Security Text result with the appropriate diagnosis when a text field is expected but not successfully recognized.
- Updated the neural network (NN) for the document occlusion check with greater accuracy.
- Updated the neural network (NN) for electronic device replay checks with greater accuracy.
- Fixed exception that might appear during the OVI check.
Enums
- Added new
eCheckDiagnose
enum member:chd_SecurityTextIsAbsent
- Added new
eRPRM_DeviceAdditionalFeatures
enum member:RPRM_DeviceAdditionalFeature_CardTimeout
Other
- Fixed issue where
sessionLogFolder
returned empty if RFID chip processing was started without prior document processing. - Fixed issue with returning the log path on Android 16.
- Fixed issue with glares detection in scenarios
MrzAndLocate
,Locate
, andMrzOrLocate
, that prevented from correct hint messages display. - Stability improvements.
Core RFID SDK
- Improved RFID chip data reprocessing to enhance security and make it less dependent on the client side, ensuring reprocessing regardless of the version in which the data was collected.
- Added the shared connectivity mode that allows multiple applications to use the same RFID reader in parallel–use the
RFID_Command_Set_DeviceConnectionMode
setting. - Fixed issue with the incorrect selection of the CSCA-Link certificate.
Enums
- Added new
eRFID_Commands
enum member:RFID_Command_Set_DeviceConnectionMode
Hardware Support
- Significantly improved performance and accuracy of document processing when working with Regula mobile devices, such as 1120 and 7310, in the
FullAuth
scenario. Improvements include better exposure adjustment depending on the environment condition and documents’ database description, UV images normalization for higher-quality results, automatic adaptation to any IR camera’s position and field of view, and environment analysis with corresponding messages (“too dark” or “too bright”). - Fixed issue that prevented a document from being extracted from the device Regula 72x3 when the Document Reader SDK was forcibly closed during operation.
- Fixed issue with Desko devices' correct scanning support.
Mobile API
- Added capability to send uncropped images to Regula Web Service when the Online Processing feature is used–they will be sent for processing if
returnUncroppedImage
is set to true. - Changed
processParam.faceApiParams.groupIds
data type fromint[]
tostring[]
. - Removed the
processParam.sessionLogFolder
field. Use theDocumentReaderResults.transactionInfo.sessionLogFolder
in the processing results instead. - Fixed issue with the
getTextFieldValueByType(fieldType: FieldType)
method retrieving the value from an incorrect source. - Updated sample projects on GitHub to align with the latest changes.
- Stability and security improvements.
Android
- Added capability to adjust the video file size saved after document processing–use the
videoRecordingSizeDownscaleFactor
setting. - Significantly improved logic of processing the
FullAuth
scenario with Regula mobile devices, such as 1120 and 7310. - The use of the Capture button has been disabled in the
FullAuth
scenario when working with Regula 1120 or Regula 7310 devices. - Fixed issue with the incorrect behavior of the back button on Android 16.
- Fixed issue where the camera frame color did not update correctly when a document was detected in multipage processing mode.
- Fixed issue with multipage animation that occurred when the camera frame size was set smaller than the default size.
iOS
- Fixed issue with incorrect display of the document liveness animation.
- Fixed issue with inactive RFID reading screen on iOS 26 that could not be closed if the device was locked beforehand.
Hybrid
- Updated
targetSdk
andcompileSdk
versions to36
for Android. - Async functions that did not actually require asynchronous execution are converted into regular functions in Flutter.
- Fixed issue with
textFieldValueByTypeSource()
that returned incorrect value in iOS.
COM API
- Added
RFIDConnectDeviceInSharedMode
property that allows multiple applications to use the RFID reader in parallel.
Document Reader Application
- Added "Personal number", "Optional data" and "Document description" fields to the list of processed documents.
- Fixed issue with incorrect reset to default value for the "Use external software for live capture" setting.
- Fixed issue with clearing barcode and MRZ processing results in video detection when the "Quick boarding pass processing" and "Quick MRZ processing" settings were enabled and “Clear results period” was set.
Web Components
- Added the light version of Web Components that allows documents capturing only—
"BoundsLite"
mode.
UI Components
- Updated the portrait comparison results visualization for cases when the Face SDK Web API is unavailable.
Web Service API
- Fixed issue with
listIdx
that wasNone
due to incorrect handling of Core results. - Updated dependencies to the latest version to fix security breaches.
Clients
- Added
Handwritten
value to theImageQualityCheck
enumeration. - Added the
BarcodePositionResult
scheme that is used for barcode position results. - Changed
processParam.faceApi.group_ids
data type fromint[]
tostring[]
. - Fixed type-related issues.
- Fixed issue in the
deserialize_to_recognition_response()
function in the Python client.
Licensing
- Fixed an issue with Face Match and Search when the Integration with Face SDK is activated and two separate licenses are used—one for the Face SDK Web API and another for the Document Reader SDK Web API, each with the corresponding capabilities. This issue does not occur when the single license is used on the Document Reader SDK side that includes capabilities from both products.
PKD
- ICAO PKD Master List version 323 included.
Database
- Updated multiple document templates to include OVI, MLI, Contact chip, Holograms, Extended MRZ, Extended OCR, Photo embedding type, Patterns checks where appropriate.
- Includes 254 countries and territories, 15753 documents.
- Added 211 new documents.
For details, see the Supported documents list.