Is there any way we can find the confidence score of the detected landmarks?
Like in face Detection, we get confidence score which tells about its accuracy, similarly is it possible in STAR Loss Model.
In STAR, you could consider use the $\sqrt{\lambda_1 \lambda_2}$ and $\lambda_1 \/ \lambda_2$ to represent the confidence and ambiguity of prediction, respectively.
Is there any way we can find the confidence score of the detected landmarks? Like in face Detection, we get confidence score which tells about its accuracy, similarly is it possible in STAR Loss Model.