search

aim-uofa / AdelaiDepth

This repo contains the projects: 'Virtual Normal', 'DiverseDepth', and '3D Scene Shape'. They aim to solve the monocular depth estimation, 3D scene reconstruction from single image problems.
Creative Commons Zero v1.0 Universal
1.07k stars 144 forks source link

training code #18

Closed caigx111 closed 2 years ago

caigx111 commented 3 years ago

Hi, Could you please provide the training code?

ttsesm commented 3 years ago

@YvanYin any update regarding publishing the training code, I would also be interested to that as quite a few my guess is in the community.

Thanks

YvanYin commented 3 years ago

Hi, I will discuss with my Adobe coauthors about traning code. There are some IP issues. I want to release them.

ttsesm commented 3 years ago

Thanks @YvanYin, looking forward to an update asap.

ttsesm commented 3 years ago

@YvanYin any update on this?

elliestath commented 2 years ago

hey, great paper! We are interested in training on our own data, will the code be eventually available? Pre-trained weights would also be very helpful.

YvanYin commented 2 years ago

I have rele

elliestath commented 2 years ago

Hey, cool, thank you so much, will test it asap. I saw it's the code on virtual normal and diverse depth projects. Is there any plan to release also the respective code for "Learning to recover 3D Scene Shape from a Single Image"?

YvanYin commented 2 years ago

Hey, most of the training code of LeReS is the same as DiverseDepth except for training data and some losses. I am still solving the IP issues, you know:) I will try to release them ASAP.

ayman-saleh commented 2 years ago

Are the differences in training data specific to the PCM, or is the DPM also trained on a separate dataset relative to DiverseDepth?

guangkaixu commented 2 years ago

@caigx111 @ttsesm @elliestath @ayman-saleh Hi, the training code of LeReS is released today.