Pith. sign in

REVIEW 2 cited by

ODIN: A Single Model for 2D and 3D Segmentation

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2401.02416 v3 pith:VP3E5RQS submitted 2024-01-04 cs.CV cs.AIcs.LGcs.RO

classification cs.CVcs.AIcs.LGcs.RO
keywords pointimagesmodelsegmentationcloudsodinperformancestate-of-the-art
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

State-of-the-art models on contemporary 3D segmentation benchmarks like ScanNet consume and label dataset-provided 3D point clouds, obtained through post processing of sensed multiview RGB-D images. They are typically trained in-domain, forego large-scale 2D pre-training and outperform alternatives that featurize the posed RGB-D multiview images instead. The gap in performance between methods that consume posed images versus post-processed 3D point clouds has fueled the belief that 2D and 3D perception require distinct model architectures. In this paper, we challenge this view and propose ODIN (Omni-Dimensional INstance segmentation), a model that can segment and label both 2D RGB images and 3D point clouds, using a transformer architecture that alternates between 2D within-view and 3D cross-view information fusion. Our model differentiates 2D and 3D feature operations through the positional encodings of the tokens involved, which capture pixel coordinates for 2D patch tokens and 3D coordinates for 3D feature tokens. ODIN achieves state-of-the-art performance on ScanNet200, Matterport3D and AI2THOR 3D instance segmentation benchmarks, and competitive performance on ScanNet, S3DIS and COCO. It outperforms all previous works by a wide margin when the sensed 3D point cloud is used in place of the point cloud sampled from 3D mesh. When used as the 3D perception engine in an instructable embodied agent architecture, it sets a new state-of-the-art on the TEACh action-from-dialogue benchmark. Our code and checkpoints can be found at the project website (https://odin-seg.github.io).

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. The Type II SN 2025pht in NGC 1637: A Red Supergiant with Carbon-rich Circumstellar Dust as the First JWST Detection of a Supernova Progenitor Star

    astro-ph.HE 2025-08 reject novelty 8.0 of 10

    SN 2025pht's progenitor is claimed to be a log(L/Lsun)=5.0 red supergiant with Av=5.3 circumstellar extinction detected in pre-explosion HST and JWST imaging, the first JWST counterpart to a supernova progenitor.

  2. Locate 3D: Real-World Object Localization via Self-Supervised Learning in 3D

    cs.CV 2025-04 conditional novelty 7.0 of 10

    LOCATE 3D localizes objects named in referring expressions directly from posed RGB-D frames, achieving 61.7% top-1 accuracy at IoU 25 on SR3D/NR3D/ScanRefer and 63.7% when trained with a new 130K-annotation dataset.

Pith tools