Platform requirements

Camera Resolution:

  • Minimum pixel density of ~30 pixels/inch (~12 pixels/cm) applied to the face
  • In a mobile phone "selfie portrait" scenario, this converts to an image resolution of 480p (i.e. 640 x 480, 4:3 aspect)

Camera Frame Rate:

  • Minimum: 25 frames per second
  • Recommended: 30 frames per second
  • Accurate frame timestamps (in milliseconds) are required

Face Tracker:

  • External face tracker that provides MPEG-4 FBA feature points
  • Anura Core SDK includes support for MediaPipe FaceMesh by default. Support for visage|SDK FaceTrack is also included.

Supported CPU Architectures:

  • ARMv8-A (64-bit): Android and iOS

Supported OS bindings/wrappers:

  • Android: Java and Kotlin
  • iOS: Swift and Objective-C
  • Cross-Platform: Please refer to the Cross-Platform Apps section for more information

OS Specific Requirements:

Android

  • OS Version: Android 7.1 (API level 25) and later

  • Camera2 API: The platform must support Camera2 API, specifically with FULL or LEVEL_3 support. Platforms that only have Camera2 API LEGACY are not supported by Anura Core SDK. While it may still run, measurement performance cannot be guaranteed.

    For more information on Camera APIs on Android, please refer to these online resources:

iOS

  • Minimum OS Version: iOS 13.0 and later
  • iPadOS 13.0 and later are also supported

Platform Performance Guidelines:

  • Comparable with Snapdragon 632 or better for Android
  • Comparable with Apple A10 or better for iOS
  • Minimal RAM footprint (6x frame buffer overhead)

Internet connectivity requirements

The payload data chunks are not very heavy (mostly less than 500 KB). Any stable internet connection which can handle such payloads is expected to work fine.