Context-synthesis mask generation during training

vt-vl-lab / 3d-photo-inpainting

[CVPR 2020] 3D Photography using Context-aware Layered Depth Inpainting

https://shihmengli.github.io/3D-Photo-Inpainting/

Other

6.9k stars 1.11k forks source link

Context-synthesis mask generation during training #118

Open surajsy opened 3 years ago

surajsy commented 3 years ago

I want to train the model on my custom dataset.

It is mentioned in the paper that context-synthesis masks are generated using a flood-fill like algorithm. Could you please share the relevant code or point me to some repository which does this?

Also, is there any post-processing of the images required during training, after the model output, like in the testing code given?

christina284 commented 7 months ago

did you find any solution to this?