AI Paper Atlas

Research map / 3D vision / Depth and stereo

NDDepth: Normal-Distance Assisted Monocular Depth Estimation and Completion

Shuwei Shao, Zhongcai Pei, Weihai Chen, Peter C. Y. Chen, Zhengguo G. Li

TPAMI 2024 · 3D vision › Depth and stereo · approach: Self-supervised monocular depth · 34 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 normal-distance head instead of depth head (ablation)NDDepth · NYU-Depth-v2 depth estimationAbs Rel: 0.095 -> 0.092Improved
add plane-aware consistency constraint (ablation)NDDepth ND head · NYU-Depth-v2 depth estimationAbs Rel: 0.092 -> 0.089Improved
add depth head alongside normal-distance head (ablation)NDDepth · NYU-Depth-v2 depth estimationAbs Rel: 0.089 -> 0.088Improved
add contrastive iterative refinement module (ablation)NDDepth · NYU-Depth-v2 depth estimationAbs Rel: 0.088 -> 0.087Improved
use normal-distance head instead of depth head (ablation)NDDepth · NYU-Depth-v2 depth completionRMSE: 0.098 -> 0.086Improved
add SPN refinement to depth head (ablation)depth head · NYU-Depth-v2 depth completionRMSE: 0.098 -> 0.090Improved
add SPN refinement to normal-distance head (ablation)NDDepth ND head · NYU-Depth-v2 depth completionRMSE: 0.086 -> 0.084Improved
add plane-aware consistency constraint (ablation)NDDepth ND head + SPN · NYU-Depth-v2 depth completionRMSE: 0.084 -> 0.082Improved
add depth head to ND head (ablation)NDDepth ND head + SPN + PC · NYU-Depth-v2 depth completionRMSE: 0.082 -> 0.081Improved
use normal-distance head instead of depth head in planar regions (ablation)NDDepth · NYU-Depth-v2 planar regionsRMSE: 0.316 -> 0.301Improved
use normal-distance head instead of depth head in non-planar regions (ablation)NDDepth · NYU-Depth-v2 non-planar regionsRMSE: 0.390 -> 0.394Hurt
NDDepth vs PixelFormer on zero-shot transferNDDepth · SUN RGB-D (zero-shot)Abs Rel: 0.144 -> 0.137Improved

Later papers that compared with this method

2 later accepted papers report NDDepth as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

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