Release Notes
Have a feature request? Contact us
1.3.0 - 2021/09/28
Breaking changes
The default model files have been renamed in order to allow for multiple models to choose from. Please see the Get Started section for an overview of the new file structure, as well as the updated API configuration options for details on how to configure model usage via the
faceDetectionModel
property. The new names for the files are:assets/express-model-shard1.xml
assets/express-model-weights_manifest.json
Changes & additions
Added a more accurate, but larger model for face detection that can optionally be used instead of the default express model. This can be configured via the
faceDetectionModel
property on the API configuration options.Added a strongly typed
ClientInfo
class that is now returned by thegetClientInfo
call.Added more detailed information to be returned from
getClientInfo
.Added custom header configuration for authentication (
authHeaders
) and passive liveness (passiveLivenessHeaders
) API calls that can be set using the API configuration options.Changed the default blur text to "Image too blurry" (see
messageBlurry
under the API configuration options).Changed the default message when no face is detected to "Scanning for face..." (see
messageNoFace
under the API configuration options).Significantly reduced the default face detection score threshold from 0.5 to 0.15 (see
faceDetectionScoreThreshold
under the API configuration options).Slightly increased the default blur threshold from 5.0 to 5.5 (see
blurThreshold
under the API configuration options).
Bug fixes
Fixed a memory leak that occurred while the video feed was being processed.
Fixed a bug where resizing while the video feed is already open would not properly adjust processing areas.
Fixed a bug on iPad where the video feed was blank.
Fixed a bug on iPad where the video cutout would sometimes not re-render correctly on resize or orientation change.
Fixed some exceptions not being handled properly.
Fixed media stream remaining open for longer than necessary after a photo was successfully taken.
Fixed styling being applied to some elements too strictly (would have needed to use !important to style said elements before).
1.2.0 - 2021/09/10
Breaking changes
Changed all
onBeforeWebcam
events toonBeforeLiveness
.
Changes & additions
Inverted the video feed when a frontal camera is being used in order to improve the alignment process.
Added typings for Typescript. These can be found under the
src
folder.The library no longer requires
sybrin.biometrics.css
to be included and referenced separately for default component styling to take effect. You may still apply overrides using your own stylesheet by referencing a custom .css file after the library.Added a read-only mode to the passive component which disables all events.
Added an option to set request headers for asset retrieval requests.
Bug fixes
Fixed a bug where an error occurred if the SDK was included multiple times.
Fixed a bug where the
correlationId
property was required to be passed down to the passive component.
1.1.0 - 2021/08/20
Changes & additions
Added blur detection as well as a configuration option to set the blur threshold.
Added lighting detection as well as configuration options to set values and thresholds to use when checking for image overexposure or underexposure.
Added facial distance detection as well a configuration option to set the distance threshold.
Removed hold delay before image is taken as new quality detection negates the need for it. With this, also removed
holdDelay
,messageHold
andmessageAbort
configuration options as they are no longer used.Added version info retrieval functionality.
Added client info retrieval functionality.
Added instructions page to the passive component in order to pre-emptively give the user some guidelines on how to ensure a viable selfie.
Added localization options to passive component initialization for changing guideline instruction texts.
Added
useModal
configuration option to the passive component which enables launching the video feed in a modal.Added a spinner that appears when a selfie is being processed during passive liveness detection.
Slightly tweaked the face bounding box borders to fit better vertically.
Updated the demo application to make use of the newly added client info retrieval functionality and initialize the passive component to use the newly added modal functionality only when the client environment is not mobile.
Bug fixes
Ensured that the video stream always closes when the passive component is destroyed.
Last updated