Research map / 3D vision / Depth and stereo
How Do Neural Networks See Depth in Single Images?
Tom van Dijk, Guido de Croon
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 changed | Where | Measured | Effect |
|---|---|---|---|
| keep both position and scale cues when varying object distance (ablation) | MonoDepth · KITTI inserted-car test | relative distance estimate: position and scale: distance estimated correctly; position only: slightly overestimated, larger SD; scale only: no chang | Improved |
| add apparent-size variation on top of vertical position (ablation) | MonoDepth · KITTI inserted-car test | distance estimate: scale-only variation gives no change in estimated distance | No clear effect |
| replace colors with grayscale (ablation) | MonoDepth · KITTI test | Abs Rel: 0.124 -> 0.130 | Hurt |
| replace colors with false colors (ablation) | MonoDepth · KITTI test | Abs Rel: 0.124 -> 0.128 | Hurt |
| replace image with semantic rgb (ablation) | MonoDepth · KITTI test | Abs Rel: 0.124 -> 0.192 | Hurt |
| replace objects with class-average colors (remove texture) (ablation) | MonoDepth · KITTI test | Abs Rel: 0.124 -> 0.244 | Hurt |
| replace objects with class-average colors (remove texture) (ablation) | MonoDepth · KITTI test | delta<1.25: 0.841 -> 0.691 | Hurt |
| add a dark shadow at the bottom of unfamiliar pasted objects (ablation) | MonoDepth · KITTI inserted objects | object detection in depth map | Improved |
| keep bottom and side edges of the car (remove center) (ablation) | MonoDepth · KITTI inserted car | detection in depth map | No clear effect |
| keep the bottom edge of the car (ablation) | MonoDepth · KITTI inserted car | detection in depth map | Improved |
| crop images to emulate camera pitch (ablation) | MonoDepth · KITTI cropped images | horizon shift in depth map: regression coefficient of 0.71 | Mixed |
Closest papers by meaning
- Visualization of Convolutional Neural Networks for Monocular Depth Estimation (ICCV 2019 · similarity 0.90)
- Depth from a Single Image by Harmonizing Overcomplete Local Network Predictions (NeurIPS 2016 · similarity 0.89)
- Improving Monocular Depth Estimation by Leveraging Structural Awareness and Complementary Datasets (ECCV 2020 · similarity 0.88)
- A Study on the Generality of Neural Network Structures for Monocular Depth Estimation (TPAMI 2023 · similarity 0.88)
- Unsupervised CNN for Single View Depth Estimation: Geometry to the Rescue (ECCV 2016 · similarity 0.87)
- DepthCues: Evaluating Monocular Depth Perception in Large Vision Models (CVPR 2025 · 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.
