Skip to content

Architecture

Jump to bottom
Aymeric edited this page Jan 13, 2015 · 26 revisions

The official documentation has been moved to http://wiki.sarah.encausse.net/

.
.
.
.
.
.
.
.

WSRKinect - WSRMicro (Client)

SARAH v3.0 introduce a big refactoring to scale from 1 to N input/output.

Architecture v3.x

Speech Recognition

All available audio input device are wrapped by WSRSpeechEngine. Here is a list of available devices:

  • Default: Default microphone
  • RTP: RTP inbout stream
  • File: Audio file recognizer
  • Kinect: All available Kinect sensors

When WSRMicro use default audio device, WSRKinect will initialize all Kinect. See also Speech API

Kinect Frame

All available Kinect are wrapped by a KinectSensor. A Kinect Sensor dispatch Color (640x480x30fps), Depth (320x240x30fps), Skeleton Frames to asynchronous Tasks.

Here is a list of Tasks:

  • WSRDepth: Perform Motion detection
  • WSRColor: Compute prominent color
  • WSRQRCode: Seek for QRCode
  • WSRGesture: Gesture recognition
    • Compute Height
    • Locate Head
  • WSRFace: Face detection and Face Recognition
    • Track Face and Mood
  • WSRCamera: Debug WPF UI

The WSRWebSocket works with first available Kinect sensor. See also Kinect API.

WSRNodeJS (Server)

Clone this wiki locally