AI Paper Atlas

Research map / 3D vision / Depth and stereo

I Can't Believe It's Not Scene Flow!

Ishan Khatri, Kyle Vedder, Neehar Peri, Deva Ramanan, James Hays

ECCV 2024 · 3D vision › Depth and stereo · approach: Learned optical and scene flow · 10 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use TrackFlow (detector plus tracker) instead of ZeroFlow XL 5xscene flow · Argoverse 2 test (Threeway EPE)Threeway EPE: 0.0015m reductionImproved
use TrackFlow instead of DeFlow on pedestrian motionscene flow · Argoverse 2 test PEDESTRIANDynamic Normalized EPE: beats DeFlow by more than 20%, 1.5x improvementImproved
TrackFlow vs DeFlow on car flowscene flow · Argoverse 2 test CARDynamic Normalized EPEHurt
replace LE3DE2E detector with BEVFusion (2% lower mAP) (ablation)TrackFlow · Argoverse 2 testDynamic Normalized EPE: +0.117944 to +0.224830Hurt
lower detector confidence threshold to raise recall (0.4 -> 0.2) (ablation)TrackFlow BEVFusion · Argoverse 2 testmean Dynamic Norm EPE: 0.8176 -> 0.4643Improved
lower confidence threshold further (0.2 -> 0.1) (ablation)TrackFlow BEVFusion · Argoverse 2 testmean Dynamic Norm EPE: 0.4643 -> 0.4816Hurt
evaluate with class-aware speed-normalized Bucket Normalized EPE instead of Threeway EPEscene flow evaluation · Argoverse 2 testmetric discriminativenessImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy