AI Paper Atlas

Research map / 3D vision / Depth and stereo

SfM-TTR: Using Structure From Motion for Test-Time Refinement of Single-View Depth Networks

Sergio Izquierdo, Javier Civera

CVPR 2023 · 3D vision › Depth and stereo · approach: Dense depth prediction and completion · 11 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 SfM sparse depths for test-time refinementAdaBins · KITTI Eigen test split, new ground truth, no croppingAbs Rel: 0.072 -> 0.060Improved
Use SfM sparse depths for test-time refinementManyDepth · KITTI Eigen test split, new ground truth, no croppingRMSE: 3.116 -> 2.648Improved
Use SfM sparse depths for test-time refinementCADepth · KITTI Eigen test split, new ground truth, no croppingAbs Rel: 0.078 -> 0.068Improved
Use SfM sparse depths for test-time refinementDIFFNet · KITTI Eigen test split, new ground truth, no croppingRMSE: 3.230 -> 2.600Improved
Replace photometric refinement with SfM-TTRCADepth · KITTI Eigen test split, new ground truth, no croppingAbs Rel: 0.088 -> 0.068Improved
Replace photometric refinement with SfM-TTRManyDepth · KITTI Eigen test split, new ground truth, no croppingRMSE: 3.034 -> 2.648Improved
Use SfM-TTR instead of no refinementManyDepth · KITTI Eigen test split, new ground truth, no croppingAbs Rel: 0.064 -> 0.057; Ph-TTR: 0.064 -> 0.056Mixed
Use median scale alignment (ablation)AdaBins + SfM-TTR · KITTI alignment ablationAbs Rel: 0.060 -> 0.074Hurt
Use strict RANSAC inlier sets rather than relaxed supervision inliers (ablation)AdaBins + SfM-TTR · KITTI alignment ablationAbs Rel: 0.060 -> 0.065Hurt
Use unweighted least-squares scale alignment (ablation)AdaBins + SfM-TTR · KITTI alignment ablationAbs Rel: 0.060 -> 0.064Hurt
Set reconstruction reliability weights to 1 (ablation)AdaBins + SfM-TTR · KITTI alignment ablationAbs Rel: 0.060 -> 0.062Hurt
Fine-tune only the encoder rather than the full network (ablation)AdaBins + SfM-TTR · KITTI encoder versus full-network refinementAbs Rel: 0.062 -> 0.060Improved

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