Pith. sign in

REVIEW 2 cited by

High Resolution Multi-Scale RAFT (Robust Vision Challenge 2022)

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 2210.16900 v1 pith:SIJR5SSI submitted 2022-10-30 cs.CV

High Resolution Multi-Scale RAFT (Robust Vision Challenge 2022)

classification cs.CV
keywords approachchallengeflowms-raftresolutionrobustvisionfirst
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

In this report, we present our optical flow approach, MS-RAFT+, that won the Robust Vision Challenge 2022. It is based on the MS-RAFT method, which successfully integrates several multi-scale concepts into single-scale RAFT. Our approach extends this method by exploiting an additional finer scale for estimating the flow, which is made feasible by on-demand cost computation. This way, it can not only operate at half the original resolution, but also use MS-RAFT's shared convex upsampler to obtain full resolution flow. Moreover, our approach relies on an adjusted fine-tuning scheme during training. This in turn aims at improving the generalization across benchmarks. Among all participating methods in the Robust Vision Challenge, our approach ranks first on VIPER and second on KITTI, Sintel, and Middlebury, resulting in the first place of the overall ranking.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

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

  1. Motion-guided sparse correction enables expert-quality point tracking across diverse microscopy regimes

    cs.CV 2026-05 unverdicted novelty 7.0

    RIPPLE recasts annotation as sparse correction on motion-guided trajectories, matching exhaustive manual quality while cutting clicks by 3-25x on five diverse microscopy datasets from jellyfish and sperm.

  2. CaRLi-V: Camera-RADAR-LiDAR Point-Wise 3D Velocity Estimation

    cs.RO 2025-11 unverdicted novelty 7.0

    CaRLi-V fuses RADAR velocity cube, camera optical flow, and LiDAR ranges in a closed-form solution to produce dense point-wise 3D velocity estimates that outperform scene flow methods on a custom dataset.