Open periscopiojoe opened 7 years ago
Hi,
Yes, you should be able to train a generic model for landmark tracking of other non-rigid shapes, such as necks or shoulders, although as I have never done this I don't know how accurate it will be. As Menpo implements a number of algorithms, and as this is a different task I can't tell which one will work better (although I am biased to thinking that CLNF would be better).
The codebase is Matlab (for training and inference) and C++ (for fast inference), so you would not be able to do it in Python, unless you port the code yourself.
Hello Tadas, congrats for your work. Is it possible to train a custom model with other landmarks not contained into the face? for exemple: neck or shoulder. I got it with Menpo, but I think that patch experts and PDM should work better, is it? And could I do it with Python? Thanks