Hacker News new | ask | show | jobs
Announcing the Matterport3D Dataset of Labeled 3D Spaces for Machine Learning (medium.com)
13 points by llebttam 3199 days ago
1 comments

This is rad! Got any ideas of stuff that can be done with this using Keras?
Lots of things! Since 3D convolutional networks are very limited in their maximum resolution, most of the interesting things you can do involve learning on RGB+D images via a 2D CNN. A lot of tasks on images (segmentation, identification etc) are easier when you have even partial depth data as input.
If you didn't see the link to the paper, it's here: https://arxiv.org/pdf/1709.06158.pdf A wide range of usecases are discussed there.