Research map / 3D vision / Depth and stereo
Generating and Exploiting Probabilistic Monocular Depth Estimates
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| use distributional outputs from the shared monocular model for depth estimation | NYUv2 monocular depth estimation · NYUv2 test | rms: 0.545 -> 0.512 | Helped |
| combine distributional monocular depth with horizontal-line measurements | Depth un-cropping · NYUv2 test, horizontal line | rms: 0.482 -> 0.431 | Helped |
| combine distributional monocular depth with sparse depth measurements | Depth completion · NYUv2 test, arbitrary sparse measurements | rms: 0.399 -> 0.359 | Helped |
| use distributional predictions rather than only the mean depth map | Pairwise ordinal depth prediction · NYUv2 test | WKDR: 30.2% -> 27.1% | Helped |
| select measurement locations using the variance map instead of random locations | Depth completion · NYUv2 test, 20 measurements | rms: 0.359 -> 0.331 | Helped |
| replace the conditional VAE with a conditional GAN (ablation) | Probabilistic monocular depth model · NYUv2 validation | S→D: 0.377 -> 0.428 | Hurt |
| reduce patch size to p=17 and stride to s=2 (ablation) | C-VAE · NYUv2 validation, depth completion from 100 measurements | S→D: 0.377 -> 0.413 | Hurt |
| increase patch size to p=65 and stride to s=8 (ablation) | C-VAE · NYUv2 validation, depth completion from 100 measurements | S→D: 0.377 -> 0.389 | Hurt |
| increase stride to s=8, reducing patch overlap (ablation) | C-VAE with p=33 · NYUv2 validation, depth completion from 100 measurements | S→D: 0.377 -> 0.396 | Hurt |
| increase stride to s=16, reducing patch overlap (ablation) | C-VAE with p=33 · NYUv2 validation, depth completion from 100 measurements | S→D: 0.377 -> 0.405 | Hurt |
| increase stride to s=32, reducing patch overlap (ablation) | C-VAE with p=33 · NYUv2 validation, depth completion from 100 measurements | S→D: 0.377 -> 0.436 | Hurt |
| reduce patch size to p=17 and stride to s=2 (ablation) | C-VAE · NYUv2 validation, oracle sample selection | Oracle: 0.323 -> 0.263 | Helped |
Closest papers by meaning
- DME: Unveiling the Bias for Better Generalized Monocular Depth Estimation (AAAI 2024 · similarity 0.89)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.88)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.88)
- DepthCues: Evaluating Monocular Depth Perception in Large Vision Models (CVPR 2025 · similarity 0.87)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · similarity 0.87)
- DepthFM: Fast Generative Monocular Depth Estimation with Flow Matching (AAAI 2025 · similarity 0.87)
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.