Skip to content

Opening book details…

About this document

Scalable Zero-Shot 3D Part Segmentation by scribd.sitetester is a document available to read on EtoBox.

The document introduces SAMPart3D, a scalable zero-shot 3D part segmentation framework that operates without preset part labels, addressing challenges in segmenting complex 3D objects. It utilizes a two-stage 2D-to-3D distillation process to learn from large-scale unlabeled 3D data, while handling semantic and granularity ambiguities. Additionally, it presents the PartObjaverse-Tiny dataset, which includes detailed annotations for 200 complex 3D objects, and demonstrates superior performance compared to exi

Author
scribd.sitetester
Language
EN