AI Paper Atlas

Research map / 3D vision / Depth and stereo

IGEV++: Iterative Multi-Range Geometry Encoding Volumes for Stereo Matching

Gangwei Xu, Xianqi Wang, Zhaoxing Zhang, Junda Cheng, Chunyuan Liao, Xin Yang

TPAMI 2025 · 3D vision › Depth and stereo · approach: Cost-volume multi-view stereo · 112 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
add single-range geometry encoding volume (GEV) (ablation)RAFT-Stereo baseline · Scene Flow Disp<192pxEPE: 0.54 -> 0.46Improved
replace GEV with multi-range GEV (MGEV) (ablation)IGEV · Scene Flow Disp<768pxEPE: 0.84 -> 0.74Improved
add adaptive patch matching (APM) (ablation)MGEV · Scene Flow Disp<768pxEPE: 0.74 -> 0.71Improved
add selective geometry feature fusion (SGFF) (ablation)MGEV+APM · Scene Flow Disp<768pxEPE: 0.71 -> 0.67Improved
use only small-range GEV instead of multi-range (ablation)IGEV++ · Scene Flow Disp<768pxEPE: 0.67 -> 1.06Hurt
use only large-range GEV instead of multi-range (ablation)IGEV++ · Scene Flow Disp<192pxEPE: 0.43 -> 0.54Hurt
use only medium-range GEV instead of multi-range (ablation)IGEV++ · Scene Flow Disp<384pxEPE: 0.56 -> 0.67Hurt
replace all-pairs correlation with MGEV, at 1 iterationRAFT-Stereo · Scene Flow Disp<768pxEPE: 3.58 -> 0.97Improved
increase iterations from 16 to 32IGEV++ · Scene Flow Disp<768pxEPE: 0.67 -> 0.67No clear effect
extend cost volume to 768px disparity rangePCWNet · Scene FlowEPE: 0.78 -> 1.52Hurt
use multi-range volumes instead of large cost volumeIGEV++ vs PCWNet · Scene Flow Disp<768pxmemory (G): 11.62 -> 1.57Improved
use DepthAnythingV2-Large frozen backboneIGEV++ · KITTI 2015 testD1-all: 1.51 -> 1.43Improved

Later papers that compared with this method

6 later accepted papers report IGEV as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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