Pith. sign in

REVIEW 4 cited by

Stereo4D: Learning How Things Move in 3D from Internet Stereo Videos

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

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 2412.09621 v2 pith:2ADZGDWN submitted 2024-12-12 cs.CV

Linyi Jin , Richard Tucker , Zhengqi Li , David Fouhey , Noah Snavely , Aleksander Holynski This is my paper
classification cs.CV
keywords datamotiontrainingdynamicestimationhigh-qualityinternetlarge-scale
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Learning to understand dynamic 3D scenes from imagery is crucial for applications ranging from robotics to scene reconstruction. Yet, unlike other problems where large-scale supervised training has enabled rapid progress, directly supervising methods for recovering 3D motion remains challenging due to the fundamental difficulty of obtaining ground truth annotations. We present a system for mining high-quality 4D reconstructions from internet stereoscopic, wide-angle videos. Our system fuses and filters the outputs of camera pose estimation, stereo depth estimation, and temporal tracking methods into high-quality dynamic 3D reconstructions. We use this method to generate large-scale data in the form of world-consistent, pseudo-metric 3D point clouds with long-term motion trajectories. We demonstrate the utility of this data by training a variant of DUSt3R to predict structure and 3D motion from real-world image pairs, showing that training on our reconstructed data enables generalization to diverse real-world scenes. Project page and data at: https://stereo4d.github.io

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

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

  1. OmniX: Any-view and Any-time 4D Reconstruction via Feed-forward Trajectory Fields

    cs.CV 2026-07 conditional novelty 6.0 of 10

    A feed-forward model uses sparse dynamic tokens and a UE5-built 80K-scene dataset to reconstruct dense 4D trajectories under large camera motion, beating prior trackers on trajectory and tracking benchmarks.

  2. Syn4D: A Multiview Synthetic 4D Dataset

    cs.CV 2026-05 unverdicted novelty 6.0 of 10

    Syn4D supplies multiview synthetic dynamic scenes with dense geometric, tracking and pose ground truth that lets any pixel be unprojected to any time and camera.

  3. Advances in 4D Representation: Geometry, Motion, and Interaction

    cs.CV 2025-10 conditional novelty 4.0 of 10

    A representation-centric survey of 4D generation and reconstruction, organized by geometry, motion, and interaction, with qualitative trade-off comparisons across seven representation families.

  4. Reconstructing 4D Spatial Intelligence: A Survey

    cs.CV 2025-07 accept novelty 4.0 of 10

    A review that classifies 4D scene reconstruction methods into five progressive levels: low-level cues, scene components, dynamic scenes, interactions, and physics.

Pith tools