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
• detectCompatibility() tests for certain browser requirements that webgazer depends on checked again in begin() returns true if browser meets expectations, false otherwise setters setTracker( libraryString ) sets which facial recognition library gazer should use returns gazer object libraryString - a string corresponding to one of the following ‘clmtrackr’ ‘js_objectdetect’ ‘trackingjs’ setRegression( regressionString ) sets which regression model type gazer should use returns gazer object regressionString - a string corresponding to one of the following ‘ridge’ ‘linear’ ‘interaction’ … 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 returns gazer object callback - function(gazeData, elapsedTime) no return value gazeData - object x - predicted gaze x screen coordinate in pixels y - predicted gaze y screen coordinate in pixels covariance - 2D confidence covariance matrix elapsedTime - time in milliseconds since begin() was invoked 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