Fuxin Li
fuxinli.bsky.social
Fuxin Li
@fuxinli.bsky.social
Computer Vision, Explainable Deep Learning, Point Clouds, 3D Perception
for the instance scene completion task.

Check out the paper here: openreview.net/pdf?id=llSiI...

Project Website:
wkhademi.github.io/point_based_...

Code/Data will be available as well (still working on it)

#ICLR2025 #instance_scene_completion (3/3)
openreview.net
February 11, 2025 at 6:59 AM
removing the need of canonical object poses and incorporating scene constraints from other objects/camera to prevent collision between completed objects. Besides, we cleaned up the alignment and collision issues with existing annotations and proposed a new dataset ScanWCF (2/3)
openreview.net
February 11, 2025 at 6:59 AM
Definitely could use some UI improvements to make it more obvious!
February 1, 2025 at 6:35 AM
Hi Jason. This information is a bit obscure, but actually it is there... When the list of all papers come up, normally below the PDF there will be a line saying "CVPR 2025 Conference Submission". If the paper is withdrawn, this line will say instead "CVPR 2025 Conference Withdrawn Submission".
January 31, 2025 at 11:35 PM