AI Paper Atlas

Research map / 3D vision / Depth and stereo

Uncertainty Guided Adaptive Warping for Robust and Efficient Stereo Matching

Junpeng Jing, Jiankun Li, Pengfei Xiong, Jiangyu Liu, Shuaicheng Liu, Yichen Guo, Xin Deng, Mai Xu, Lai Jiang, Leonid Sigal

ICCV 2023 · 3D vision › Depth and stereo · approach: Learned stereo with specialized sensors · 42 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 uncertainty estimation and content-aware warping instead of bilinear warping (ablation)Lite-CREStereo++ · Middleburybad 2.0: 14.79 -> 11.86Improved
add uncertainty estimation on top of content-aware warping (ablation)Lite-CREStereo++ · KITTI2015D1-all: 2.47 -> 2.36Improved
content-aware learnable warping without uncertainty vs bilinear (ablation)Lite-CREStereo++ · Middleburybad 2.0: 14.79 -> 12.96Improved
use variance-based uncertainty instead of error-based map (ablation)Lite-CREStereo++ · Middleburybad 2.0: 12.40 -> 11.86Improved
sigmoid instead of tanh in variance uncertainty (ablation)Lite-CREStereo++ · Middleburybad 2.0: 11.98 -> 11.86Improved
GRU kernels 1x5 + 1x15 instead of 3x3 (ablation)Lite-CREStereo++ · Middleburybad 2.0: 14.21 -> 11.86; +8ms runtimeImproved
add a third 1x31 GRU kernel (ablation)Lite-CREStereo++ · Middleburybad 2.0: 11.86 -> 11.98Hurt
CREStereo++ vs CREStereo in cross-domain generalization from Scene FlowCREStereo · ETH3Dbad 1.0: 5.5 -> 4.4Improved
use UGAC in lightweight model with 0.6M parametersLite-CREStereo++ · KITTI 2012 online3-all: 1.89 -> 1.82 vs HITNetImproved

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