Top Level API

Here are all the functions you call to control the operation of WebGazer.

• on include of script in body
• gazer namespace object introduced
• overwrites any other gazer data to prevent data contention
• gazer singleton namespace object
• functions
  • gazer begin()
    • begins collecting click data and activates the camera, should be called as soon as possible to improve accuracy
    • attaches window click listener
    • starts clock
    • video div attached to body
      • id = webgazerVideoFeed
      • style = invisible
    • canvas attached to body
      • id = webgazerCanvas
      • style = invisible
    • returns gazer object
  • gazer pause()
    • stops all gazer data collection and predictions
    • returns gazer object
  • gazer resume()
    • resumes all gazer data collection and predictions after a pause()
    • returns gazer object
  • gazer end()
    • stops all gazer data collection and predictions and releases all memory being held by the gazer object.
    • Accuracy may suffer if begin is called afterward, consider using pause and resume instead
    • IMPORTANT: saves global model data so that user’s returning to the page will have a pre-loaded model
    • returns gazer object
  • gazer showPredictionPoints(boolean)
    • if bool is true: shows a circle with the current best guess given by the prediction model
    • if bool is false: turns off the prediction circle
    • returns gazer object
  • boolean detectCompatibility()
    • tests for certain browser requirements that webgazer depends on
    • checked again in begin() //TODO
    • returns true if browser meets expectations, false otherwise
• setters
  • gazer setTracker( libraryString )
    • sets which facial recognition library gazer should use
    • libraryString - a string corresponding to one of the following
      • ‘clmtrackr’
      • ‘js_objectdetect’
      • ‘trackingjs’
    • returns gazer object
  • gazer setRegression( regressionString )
    • sets which regression model type gazer should use
    • regressionString - a string corresponding to one of the following
      • ‘ridge’
      • ‘linear’
      • ‘interaction’
    • returns gazer object
  • gazer setGazeListener( callback )
    • sets a callback which will be invoked every time webgazer makes a new prediction
    • only one callback can be registered at a time, so in multiple calls to setGazeListener only the last callback will remain registered
    • callback - function(gazeData, elapsedTime) //TODO this has changed to include results of all regression models
      • gazeData - object
        • x - predicted gaze x screen coordinate in pixels
        • y - predicted gaze y screen coordinate in pixels
        • covariance - 2D confidence covariance matrix //TODO
      • elapsedTime - time in milliseconds since begin() was invoked
    • returns gazer object
  • gazer clearGazeListener()
    • removes the current callback
    • use this method instead of pause/resume when data should continue to be collected to improve the accuracy
    • returns gazer object

getters getTracker() gets the current facial recognition library returns string - the string corresponding to the library getRegression() gets the current regression model type returns string - the string corresponding to the regression model getCurrentPrediction() request a prediction from gazer returns object x - predicted gaze x screen coordinate in pixels y - predicted gaze y screen coordinate in pixels covariance - 2D confidence covariance matrix