Code for generating object detection dataset

[bamps53]

Hi, thanks for sharing the interesting work! do you have a plan to release the code for generating object detection dataset? (which is described in IV. EXPERIMENTS B in https://arxiv.org/pdf/2010.11696.pdf)

[cheind]

Hey!

The code is already made public

https://github.com/cheind/pytorch-blender-dr

However, the code ist not in a very clean state. We wanted to fix this before reintegrating it into pytorch-blender, but haven't found the time yet. In particular, the relevant branches are

https://github.com/cheind/pytorch-blender-dr/tree/blexp/blender/tless

and a good model was also

https://github.com/cheind/pytorch-blender-dr/blob/camera_noise/blender/tless/

which refer to pytorch-blender branch

https://github.com/cheind/pytorch-blender/tree/feature/compositor_support

for compositor support (camera noise postprocessing).

I will get in touch with @lukasbrunnerjku for his comments.

[bamps53]

Wow, I didn't realize it! That's awesome. Thank you so much for detailed answer and support. I'll look through it:)

[cheind]

No worries, it was not meant to be found at this stage :)

[bamps53]

That's why I thanks more, you could just say "we'll release soon", but you didn't ;)

[lukasbrunnerjku]

Hi, I can clean it next weekend so you can have it as soon as possible :)

[cheind]

@bamps53, @lukasbrunnerjku just merged compositor-support alongside with some other fixes to develop. please let me know if that causes troubles for tless.

[bamps53]

Thank you so much! I'll try it out:) Thanks again for your generous support.

[cheind]

Hey @bamps53, lukas just merged a cleaned up version to master:

https://github.com/cheind/pytorch-blender-dr

Hope you find it useful