Train 3D segmentation network using simple bounding boxes with Box2Mask
Train 3D segmentation network using simple bounding boxes with Box2Mask
Semantic Instance Segmentation of 3D Scenes Through Weak Bounding Box Supervision
arXiv paper abstract https://arxiv.org/abs/2206.01203
arXiv PDF paper https://arxiv.org/pdf/2206.01203.pdf
Project page https://virtualhumans.mpi-inf.mpg.de/box2mask
Current 3D segmentation methods heavily rely on large-scale point-cloud datasets, which are notoriously laborious to annotate.
... idea is to leverage 3D bounding box labels which are easier and faster to annotate.
... show that it is possible to train dense segmentation models using only weak bounding box labels.
... method, Box2Mask ... directly votes for bounding box parameters, and a clustering method specifically tailored to bounding box votes.
This goes beyond commonly used center votes, which would not fully exploit the bounding box annotations.
... weakly supervised model attains leading performance ... also achieves 97% of the performance of fully supervised models ...
Please like and share this post if you enjoyed it using the buttons at the bottom!
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
Opmerkingen