This page contains the release notes for the different versions of the Affdex SDK(s).


Apr 2017

Linux SDK 3.3
  • Fix bug - shared state affecting initialization of multiple detectors
  • Fix bug - reduced memory usage in long running sessions
Windows SDK 3.4.1
  • Fix bug - shared state affecting initialization of multiple detectors
  • Fix bug - reduced memory usage in long running sessions

Feb 2017

Web SDK 3.2
  • Revised Nose Wrinkle and Smirk models to lower their false positive rates
  • Updated Valence model to recalibrate the low Valence range
  • Updated Expressiveness to use the new Nose Wrinkle model
  • Updated Fear, Surprise, Anger, Disgust, Sadness, and Contempt emotions to include latest expression models
Windows SDK 3.4
Linux SDK 3.2

Jan 2017

Android SDK 3.2
  • Added support for “brightness” metric to measure the face lighting condition. See documentation for [qualities] field in [Face] class
  • Revised Nose Wrinkle and Smirk models to lower their false positive rates
  • Updated Valence model to recalibrate the low Valence range
  • Updated Expressiveness to use the new Nose Wrinkle model
  • Updated Fear, Surprise, Anger, Disgust, Sadness, and Contempt emotions to include latest expression models
  • Fixed bug SDK-1053 - analytics frame records’ timestamps should follow detector-specific rules
  • Fixed bug SDK-1003 - SDK should still send start/stop analytics records for non-empty sessions when analytics are disabled
  • Fixed bug SDK-529 - native crash if Detector’s process method() is called after stop()
Windows SDK 3.3
  • Added support for “brightness” metric to measure the face lighting condition. See documentation for faceQuality field in Face class [c++], and FaceQuality property in Face class [c#]
Windows SDK 3.2
  • Revised Nose Wrinkle and Smirk models to lower their false positive rates
  • Updated Valence model to recalibrate the low Valence range
  • Updated Expressiveness to use the new Nose Wrinkle model
  • Updated Fear, Surprise, Anger, Disgust, Sadness, and Contempt emotions to include latest expression models

Nov 2016

Windows SDK 3.1.2
  • Fixed bug SDK-763 - Added the RGBA, BGRA, YUV_NV21, YUV_I420, YUV_YUY2 color formats for the enum in Frame class

Oct 2016

Unity SDK 2.3
  • Added iOS support

Sep 2016

Unity SDK 2.2.1
  • Android: Updated the libpng version used in the native library to address a security vulnerability
Android SDK 3.1.2
  • Updated the libpng version used in the native library to address a security vulnerability
Windows/Linux/Android SDK 3.1.1
  • Removed license-related methods. A license file is no longer required to use the SDK.
iOS/macOS SDK 3.1.1
  • Removed license-related methods. A license file is no longer required to use the SDK.
  • API allows developer to choose face detector’s face size configuration (SMALL / LARGE)

Aug 2016

Unity SDK 2.2
  • Added Android support.
  • Frame constructor can now accept a rotation angle of 0, 90, 180, and 270.
  • Detector.Initialize is now private because it is called automatically by Detector.Start
  • Detector.Stop has been deprecated in favor of Detector.StopDetector

Jul 2016

Windows/Linux/Android SDK 3.1
  • Improved emotion metrics. The valence includes new expressions in its calculation.
  • Added models for classifying ethnicity and estimating age range.
  • Added models for new expressions: cheek raise, dimpler, eye widen, lid tighten, lip stretch and jaw drop.
  • Fixed bug SCI-86 - Disgust false positives due to the occurrence of the smile expression.
Web SDK 3.1
  • First release of our JavaScript-based SDK to emotion-enable Web apps.

May 2016

macOS SDK 3.0.2
  • Performance optimizations, SDK-521 - remove dispatch_async() overhead from callback functions.

Mar 2016

macOS SDK 3.0.1
  • First release of our Affdex SDK for the OS X platform.
Android SDK 3.0.1
  • Switch the SDK packaging to the new Android standard packaging (aar).
iOS SDK 3.0.1
  • Fixed bug SDK-1 - iOS Framework builds v3.0 are much bigger in size than v2.0.
  • Fixed bug SDK-38 - iOS device Framework contains i386 and x86_64 architectures.
  • Fixed bug SDK-404 - didStopDetectingFace: can pass nil for the face parameter.
Linux SDK 3.0
  • First release of our Affdex SDK for the Linux platform.

Jan 2016

Windows/iOS/Android SDK 3.0
  • New API to allow tracking of multiple faces simultaneously.
  • Added a Face to Emoji mapper to allow the developer to map facial expressions of emotion to 13 emojis different emojis
  • New API that returns the most likely emoji (dominantEmoji) posed in an image
  • Added models to allow identification of gender and presence of eyeglasses or sunglasses
  • New API that allows the developer to control the face detector configuration to use in the detection
  • Added dynamic core detection to allow SDKs to make use of multiple CPU cores if available
  • New Getting Started tutorials on the Developer Portal
  • New version of the AffdexMe apps (iOS, Android and Windows):
    • Track multiple faces in realtime On Windows this feature is enabled by default, but on iOS/Android it can be toggled from the settings menu.
    • Display the metrics for Facial Expression, Emotion, Appearance and Dominant Emoji
    • Allow users to take screenshots of the camera feed with the metrics overlay
    • Allow users to use the back camera to track the metrics
    • Optimize AffdexMe (Windows) metric and camera visualization class
  • iOS SDK
    • AFDXFace class re-organized into categories expressions, emotions, emoji and appearance
    • Expressiveness renamed into Engagement
    • Head angles renamed HeadUpDown, HeadLeftRight, HeadRoll to Pitch, Yaw, Roll
    • Added API to allow embedding the license string as a variable in the app code
  • Win SDK
    • Performance improvements
    • Fixed bug: The assembly version doesn’t match the file metadata
    • Fixed bug: Calling reset() doesn’t reset the internal cached timestamps.
  • Android SDK
    • Fixed bug: the license token validation fails due to zero padding

Sep 2015

Windows/iOS/Android SDK 2.0
  • Dramatically improved accuracy with new expression models trained over a broader set of facial images.
  • Better ease of use and accessibility by standardizing models for seven emotions: Anger, Sadness, Disgust, Joy, Surprise, Fear and Contempt.
  • New facial expressions, including very nuanced expressions that are difficult even for the human eye to detect:

    Attention, Brow Furrow, Brow Raise, Inner Brow Raise, Eye Closure, Nose Wrinkle, Upper Lip Raise, Lip Suck, Lip Pucker, Lip Press, Mouth Open, Lip Corner Depressor, Chin Raise, Smirk, Smile.

  • 3D head orientation metrics

    Left/Right, Up/Down, Roll

  • Supports the development of 64-bit applications for Windows.
  • New Developer Portal with documentation and source code for sample apps, including the flagship AffdexMe demo app.