Hello.
I have been working to get the aam-library working with the updated API.
Presently, i have a training and fitting method that runs. The result however is undesirable as the model descends into the subject's eye.
This project has been a simple case of refactoring. A fresh pair of eyes would be useful at this stage.
Is anyone available to compare my code with that of the original aam-library? If successful, we will both have working AAM code in OpenCV.