AI Paper Atlas

Research map / 3D vision / Human pose estimation

Wide-Baseline Relative Camera Pose Estimation With Directional Learning

Kefan Chen, Noah Snavely, Ameesh Makadia

CVPR 2021 · 3D vision › Human pose estimation · approach: Learned camera pose estimation · 45 citations (OpenAlex)

Check 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
Represent pose components as discrete directional distributions rather than directly regressing poseDirectionNet versus 6D regression · Matterport-A, Matterport-B, InteriorNet-A, InteriorNet-Brotation and translation mean angular error: Matterport-A rotation 5.73 -> 4.30; translation 35.75 -> 16.37; Matterport-B rotation 18.23 -> 14.85; translation 39.06 Improved
Factorize rotation into spherical direction vectors instead of estimating a discrete quaternion distributionDirectionNet · Matterport-A and Matterport-Brotation mean angular error: 3.96 vs 23.32; 13.60 vs 37.09Improved
Add intermediate derotation and estimate translation in a separate second stage (ablation)DirectionNet-9D · Matterport-A, Matterport-B, InteriorNet-A, InteriorNet-Btranslation mean angular error: 22.84 -> 14.17; 29.45 -> 21.26; 19.53 -> 12.36; 31.87 -> 16.36Improved
Use SVD projection rather than partial Gram-Schmidt projection (ablation)DirectionNet rotation estimation · Matterport-A and InteriorNet-Arotation mean angular error: 4.30 -> 3.96; 2.90 -> 2.87Improved
Add distribution loss (ablation)DirectionNet-9D · Matterport-Arotation mean angular error: 14.66 -> 3.96Improved
Add direction loss (ablation)DirectionNet-9D · Matterport-Arotation mean angular error: 4.85 -> 3.96Improved
Add spread loss (ablation)DirectionNet-9D · Matterport-Arotation mean angular error: 4.68 -> 3.96Improved
Perturb the estimated rotation used for derotation during translation training (ablation)DirectionNet-T · InteriorNet and Matterport-Btranslation angular error: 4° worse on InteriorNet without perturbationImproved
Use dense MSE supervision for distributions instead of KL divergence (ablation)DirectionNet · Experiments reported in paperperformanceImproved
Train with 100% rather than 20% of KITTI training data (ablation)DirectionNet-9D · KITTIrotation and translation mean angular error: 10.50 -> 9.19; 26.74 -> 19.36Improved
Use SIFT+LMedS rather than SIFT+RANSAC (ablation)SIFT feature-based pose estimation · Matterport and InteriorNet datasetsrotation and translation angular errorMixed

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