Research map / 3D vision / Depth and stereo
Gradient-Based Uncertainty for Monocular Depth Estimation
Julia Hornauer, Vasileios Belagiannis
This paper tried a common change: Add dropout
add inference-time dropout uncertainty · KITTI stereo supervision · AUSE RMSE: 2.523 -> 3.366 Hurt
What 795 accepted papers report for this change
- 549 improved
- 87 no clear effect
- 12 mixed
- 147 hurt
- Don't Drop Dropout: Optimizing Layer Sparsity for Efficient LLM Training and Inference ICML 2026 · apply layer dropout to whole transformer blocks instead of attention and FFN sub-blocks separately · 20 TPP val loss, 271M · val loss: 2.377 -> 2.367Improved
- Quantifying the Uncertainty of Foundation Models with Singular Value Ensembles ICML 2026 · use MC Dropout · vision and SST-2 · accuracy: 86.3 -> 76.6 on Flowers102Hurt
- Motion Planning in Compressed Representation Spaces ICML 2026 · add nested dropout during training · WOMD guided generation · goal FDE: 4.145 -> 1.104Improved
See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.
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 |
|---|---|---|---|
| add gradient-based uncertainty | Monodepth2 Post · KITTI monocular supervision | AUSE RMSE: 2.864 -> 0.540 | Improved |
| add gradient-based uncertainty | Monodepth2 Self · KITTI monocular supervision | AUSE Abs Rel: 0.030 -> 0.024 | Improved |
| add gradient-based uncertainty | Monodepth2 Post · KITTI stereo supervision | AUSE RMSE: 2.523 -> 0.510 | Improved |
| add gradient-based uncertainty | Monodepth2 Self · KITTI stereo supervision | AUSE Abs Rel: 0.022 -> 0.022 | No clear effect |
| add gradient-based uncertainty | Monodepth2 Post · NYU Depth V2 | AUSE Abs Rel: 0.066 -> 0.061 | Improved |
| add gradient-based uncertainty | Monodepth2 Log · NYU Depth V2 | AUSE RMSE: 0.159 -> 0.176 | Hurt |
| add inference-time dropout uncertainty | Monodepth2 Post · KITTI stereo supervision | AUSE RMSE: 2.523 -> 3.366 | Hurt |
| add inference-time dropout uncertainty | Monodepth2 Post · KITTI monocular supervision | AUSE RMSE: 2.864 -> 0.871 | Improved |
| use flip reference loss instead of ground-truth loss (ablation) | Gradient uncertainty Post · NYU Depth V2 | AUSE Abs Rel: 0.020 -> 0.061 | Hurt |
| use flip instead of grayscale reference (ablation) | Gradient uncertainty Post · NYU Depth V2 | AUSE Abs Rel: 0.067 -> 0.061 | Improved |
| use flip instead of 20 degree rotation reference (ablation) | Gradient uncertainty Post · NYU Depth V2 | AUSE Abs Rel: 0.076 -> 0.061 | Improved |
| use flip instead of Gaussian noise reference (ablation) | Gradient uncertainty Post · NYU Depth V2 | AUSE RMSE: 0.261 -> 0.252 | Improved |
Closest papers by meaning
- Revisiting Gradient-Based Uncertainty for Monocular Depth Estimation (TPAMI 2025 · similarity 0.95)
- Uncertainty Quantification in Depth Estimation via Constrained Ordinal Regression (ECCV 2022 · similarity 0.89)
- Unsupervised Monocular Depth Estimation With Left-Right Consistency (CVPR 2017 · similarity 0.89)
- Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement (ECCV 2018 · similarity 0.88)
- Adaptive Confidence Thresholding for Monocular Depth Estimation (ICCV 2021 · similarity 0.88)
- Learning the Distribution of Errors in Stereo Matching for Joint Disparity and Uncertainty Estimation (CVPR 2023 · similarity 0.88)
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.
