Understanding the paper

[Mathilda88]

Thanks for the great job. I have some difficulty in finding if tau_sr, tau_dr, J(tau_sr), and J(tau_dr) are four outputs from one or multiple networks that are gonna be used later in some loss constraints? All of them are being predicted by a neural network or some are calculated like the one in "deformable GAN"? Any help?

[AliaksandrSiarohin]

All predicted T_sr(pk), J_sr= kp_detector(s), same for d. It is called value and jacobian.

[ParikhKadam]

Hi @AliaksandrSiarohin

I am reading your paper and trying to understand bits in the code. I am unable to understand how jacobian and is being calculated and used.

Can you please describe it in brief in a layman's term?

Thank you..