I've been trying this for weeks, I was told it would be straightforward and that if I wanted to import the face class I could simply "add that option to the recipe". But this is somehow starting to seem more difficult than I first thought. Has anyone ever managed to compile any of the extra modules (contrib) through p4a by changing the opencv recipe? If this is indeed too difficult, could I use pyjnius, as only 10 lines of code in my entire app is dedicated to detecting facial landmarks.