Research map / 3D vision / Depth and stereo
Single-Image Depth Perception in the Wild
Read the paperCheck 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 an end-to-end pixel-wise depth network trained with relative-depth supervision rather than the Zoran method | Single-image depth estimation · NYU Depth test | WKDR: 43.5% -> 35.6% | Helped |
| Train on full NYU Depth and use all available pairs (ablation) | Ours · NYU Depth test | WKDR: 35.6% -> 28.3% | Helped |
| Replace superpixel pair sampling with 3K random pairs per image (ablation) | Ours · NYU Depth test | WKDR: 35.6% -> 35.8% | No clear effect |
| Increase random point pairs per image to 12K (ablation) | Ours · NYU Depth test | WKDR: 36.1% -> 34.9% | Helped |
| Pre-train on NYU Depth and fine-tune on DIW (ablation) | Ours · Depth in the Wild test | WHDR: 22.14% -> 14.39% | Helped |
| Train only on NYU Depth instead of fine-tuning on DIW (ablation) | Ours · Depth in the Wild test | WHDR: 14.39% -> 31.31% | Hurt |
| Reduce supervised superpixel pairs from 800 to 500 per image (ablation) | Ours · NYU Depth test | WKDR: 35.6% -> 37.2% | Hurt |
| Reduce supervised superpixel pairs from 800 to 250 per image (ablation) | Ours · NYU Depth test | WKDR: 35.6% -> 38.0% | Hurt |
| Reduce supervised superpixel pairs from 800 to 100 per image (ablation) | Ours · NYU Depth test | WKDR: 35.6% -> 42.3% | Hurt |
Closest papers by meaning
- DepthCues: Evaluating Monocular Depth Perception in Large Vision Models (CVPR 2025 · similarity 0.87)
- Depth Prompting for Sensor-Agnostic Depth Estimation (CVPR 2024 · similarity 0.86)
- RSA: Resolving Scale Ambiguities in Monocular Depth Estimators through Language Descriptions (NeurIPS 2024 · similarity 0.86)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.86)
- Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data (CVPR 2024 · similarity 0.86)
- Depth Anything V2 (NeurIPS 2024 · similarity 0.86)
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.