How to improve stability of face tracking?

[hyzhak]

Shortly: How to improve stability of face tracking in general and tracking faces with in different light condition in particular?

---

Thanks for you library I love it a lot. But it's very unstable with my face face :(. Maybe it is not common because I'm bald and have a beard. But I have request from my client as well, he told that it is fairly unstable on face tracking and regularly loses tracking. And I have found that it is very hard to track face when it is lit from one side.

Could you share your ideas how can we improve this lib to fit more kind of faces? And prevent it from losing face tracking.

Here is example how emotions tracker works for my face

PS: My issue seems somehow related to issue #8

[Skylion007]

Unfortunately, CLMTrackr does not appear to be well trained on people with facial hair. You can help rectify this injustice by using CLMTools to retrain the model.