[klyrs]

Sounds like you're using machine learning & computer vision. How much diversity is in your training set? I can see obesity, skin tone, and gender playing a role in this. Or, is it entirely trained on photos of the user? Even so... I see you're looking for users' ears. Can it cope with ears hidden by hair, or chunky earrings that change daily?

[handheldmouse]

The training set used has a lot of diversity. There is no training done on frames of the user - those photos are not stored, process or transferred in any way.

We have users with long hair that covers their ears or users that wear headphones that go over their ears and head and they use the app perfectly fine. One thing that can affect the accuracy of the algorithms is the lighting in the room - poor or extremely strong lighting in some cases can lead to slight inaccuracies.

[bobobob420]

Sorry if I am not understanding but you say the photos of user are not processed but aren’t the frames from the webcam stream processed to extract data about the users posture?