search

cheind / pytorch-blender

:sweat_drops: Seamless, distributed, real-time integration of Blender into PyTorch data pipelines
MIT License
562 stars 44 forks source link

datagen other complex scenes (than cubes) #11

Closed ggaziv closed 3 years ago

ggaziv commented 3 years ago

Hi, thanks for this amazing package ! I was wondering how to go about making diverse scenes other than cubes. Is there anywhere I could download these? Specially, say I download some .blend files off the internet, how to make suitable .py files for their integration? Also, how does background diversification come into play?

Thanks ! Guy

cheind commented 3 years ago

Hey,

our latest paper https://arxiv.org/abs/2010.11696 considers object detection as an example. Here the TLESS dataset provides the CAD files of objects, the scene is a simple planar surface. We randomize objects above the plane and let physics create constellation of fallen objects. The code can be found here https://github.com/cheind/pytorch-blender-dr/tree/master/blender/tless but is not necessarily very clean at this point.