Inference without shape priors ? Does it change anything ?

[chronos-ctrl]

Hi sifei,

In which place the non-parametric priors are included as a complementary channels (as it was introduced in the paper) ?

Because as I saw in run_demo.m, the inference is done only using the aligned image channels.

Thanks for your help,

[Liusifei]

Hi Amine, Yes, you are correct. This version does not include the prior. It is a multi-objective network with RGB input only. Best, Sifei Liu 2017-05-22 5:46 GMT-07:00 Amine <notifications@github.com>:

…

Hi sifei, In which place the non-parametric priors are included as a complementary channels (as it was introduced in the paper) ? Because as I saw in run_demo.m, the inference is done only using the aligned image channels. Thanks for your help, — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#4>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AE6AI7jN500RNbsjrZqvf4r5RfvTHDEYks5r8YO3gaJpZM4NiTqy> .

-- Yours faithfully, --------------------- Sifei Liu Ph.D. candidate EECS UC Merced

[HPL123]

@aminm2a Hi, I only get some facial landmark points in the Helen dataset,and How to obtain face semantic labels to train the network?