-
Thank you for your work.
Why is the comparison data in the Per plane recall of Figure 9 in this article different from the comparison data in the Per pixel recall of Figure 4 in (PlaneNet)? For exam…
-
Hi,
It seems like the write_ply.py can only generate 3D point cloud model with a single channel, and the RGB values of each pixel are missing after the 3D remapping. I guess I might still missing som…
-
**System information**
- OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Linux Ubuntu 16.04
- TensorFlow installed from (source or binary): binary
- TensorFlow version (or github SHA if fr…
-
Hey, thanks for your excellent work.
I have tried your PlaneNet version, and wanna used that in kitti dataset, but it worked not really well.
In the paper, I see the huge potential of PlaneRCNN. An…
-
Hi!
I know it seems like a silly question but I believe it's crucial. That's because if we are talking about plane segmentation, we must have a clear idea about what a plane is. However, after seei…
-
Hi. Congratulations for your great achievement with this research!
I was wondering if have any implementation that I could perform to use the plane instance segmentations generated (that represent …
ghost updated
5 years ago
-
Hi, thank you for your wonderful work!
It seems that we can only eval on ScanNet's data which in tfrecord format.
So is there any plan to give a test code that we quickly test on single image and ge…
-
Hi, thanks for the great paper and code.
I have the code working ok, but the normals it predicts look very strange to me. For example, for a vertical surface I'd expect the y-component to be near z…
-
Thanks for opensourcing PlaneNet, Its behavior at corridor is amazing.
I'd like to train a revised PlaneNet from scratch, while the data your team provided has no planenet_scannet_train.tfrecords.
…