The ScanNet dataset by Dai et al. was designed to support research and development in the field of 3D computer vision, particularly for tasks related to scene understanding, 3D reconstruction, semantic segmentation, and object recognition. It was recently used as part of SAM2Point: Segment Any 3D as Videos in Zero-shot and Promptable Manners, by Guo et al.
http://www.scan-net.org/
SAM2Point is a preliminary exploration adapting Segment Anything Model 2 (SAM 2) for zero-shot and promptable 3D segmentation. SAM2Point interprets any 3D data as a series of multi-directional videos, and leverages SAM 2 for 3D-space segmentation, without further training or 2D-3D projection. The framework supports various prompt types, including 3D points, boxes, and masks, and can generalize across diverse scenarios, such as 3D objects, indoor scenes, outdoor environments, and raw sparse LiDAR. Demonstrations on multiple 3D datasets, e.g., Objaverse, S3DIS, ScanNet, Semantic3D, and KITTI, highlight the robust generalization capabilities of SAM2Point. To the authors' best knowledge, they present the most faithful implementation of SAM in 3D, which may serve as a starting point for future research in promptable 3D segmentation. Online Demo: https://lnkd.in/dK5rrhrf . Code: https://lnkd.in/dCU2vqQg .
I have provided an overview of the main features of this dataset, such as accessibility, suggested uses, labelled features etc. at https://lnkd.in/dCRuGHdc
CREATIVE - MIND THE FONT® & MONTEREY BAY AQUARIUM // PODCAST CO-HOST - KIRK + KURTTS DESIGN & 10 TON POD
2moThis was so much fun. Thanks for inviting me out. Great group of folks at GS1 and I hope that I have the opportunity again.