To use these API on Outsystems environment, it was necessary change the faceApi script and the resources of 'tiny face' model.

The Tiny Face Detector is a lightweight, real-time face detection model optimized for performance on low-resource environments. It provides fast and efficient face detection with lower accuracy compared to larger models, ideal for scenarios requiring speed over precision.

The functionalities are very simple:

select the searchtype : 

oneface: closest face  

allfaces: all the faces on the media

detectionscore:

percentage confidence that a face was found (for single face only) 0.1 to 1

refresh

how often should the refreshment of the search for a face ms( 1 second = 1000)

Then the faceApi will return the binary of faces finded.

There are many problems to return a binary image list directly, for that, if you use the 'allfaces' method, the system will save the images and send all after the finding ends. This is the only server side action, everything else is made on the client side.