English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Instance segmentation on 3D point clouds (3DIS) is a longstanding challengein computer vision, where state-of-the-art methods are mainly based on fullsupervision. As annotating ground truth dense instance masks is tedious andexpensive, solving 3DIS with weak supervision has become more practical. Inthis paper, we propose GaPro, a new instance segmentation for 3D point cloudsusing axis-aligned 3D bounding box supervision. Our two-step approach involvesgenerating pseudo labels from box annotations and training a 3DIS network withthe resulting labels. Additionally, we employ the self-training strategy toimprove the performance of our method further. We devise an effective GaussianProcess to generate pseudo instance masks from the bounding boxes and resolveambiguities when they overlap, resulting in pseudo instance masks with theiruncertainty values. Our experiments show that GaPro outperforms previous weaklysupervised 3D instance segmentation methods and has competitive performancecompared to state-of-the-art fully supervised ones. Furthermore, we demonstratethe robustness of our approach, where we can adapt various state-of-the-artfully supervised methods to the weak supervision task by using our pseudolabels for training. The source code and trained models are available athttps://github.com/VinAIResearch/GaPro.

GaPro: Box-Supervised 3D Point Cloud Instance Segmentation Using Gaussian Processes as Pseudo Labelers