Benjamin Ummenhofer

We are working on adding UVAtlas in #5376

The returned depth map has already been scaled with the scaling factor. It is not possible to obtain a depth map with a real world unit. In general this is...

Yes, in general it is not possible to infer the scale. The predicted depth map is scaled such that the depth values correspond to the unit translation. If you know...

Yes, this is because of the training procedure, which uses multiple stages. Using github issues for these questions is fine.

It looks like tensorflow is mapping some ops to the CPU. Unfortunately, tensorflow still does not support NCHW for some ops. Maybe you can try to strictly place ops on...

Hi @JiamingSuen , thanks for checking out our training code! 1. At the moment we use the hyperparameters as in the training code. There is probably a lot of room...

Hi, we used the original image and depth files for the rgbd-slam dataset when creating the h5 files.

We are aware of this problem but unfortunately don't have a solution for this yet. In this case it is better to not adjust the intrinsics, although the network was...

The ranges for the ground truth data are basically correct. inf for inverse depth should be avoided and the optical flow can be larger or smaller than 1.0 or -1.0...

> One issue is the flow_to_depth op always expects intrinsics to be a (N, 4) tensor. See here: https://github.com/lmb-freiburg/lmbspecialops/blob/master/src/flowtodepth.cc#L361 > > This works fine with a batch size of 1...