Research map / 3D vision / Depth and stereo
Learning Depth From Monocular Videos Using Direct Methods
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 |
|---|---|---|---|
| Add inverse-depth normalization before the loss (ablation) | Monocular-video depth training with Pose-CNN · KITTI Eigen test split (K) | abs rel: 0.213 -> 0.155 | Helped |
| Add inverse-depth normalization before the loss (ablation) | Monocular-video depth training with Pose-CNN · KITTI Eigen test split (K) | sq rel: 3.770 -> 1.193 | Helped |
| Replace Pose-CNN pose estimation with DDVO initialized at identity (ablation) | Monocular-video depth training · KITTI Eigen test split (K) | abs rel: 0.155 -> 0.159 | Hurt |
| Initialize DDVO with a pretrained Pose-CNN, then refine pose geometrically (ablation) | Monocular-video depth training · KITTI Eigen test split (K) | abs rel: 0.155 -> 0.151 | Helped |
| Initialize DDVO with a pretrained Pose-CNN, then refine pose geometrically (ablation) | Monocular-video depth training · KITTI Eigen test split (K) | RMSE: 5.613 -> 5.583 | Helped |
| Pretrain on Cityscapes before KITTI training (ablation) | Ours (Pose-CNN + DDVO) · KITTI Eigen test split | abs rel: 0.151 -> 0.148 | Helped |
| Pretrain on Cityscapes before KITTI training (ablation) | Ours (Pose-CNN + DDVO) · KITTI Eigen test split | abs rel: 0.148 vs 0.124 | Hurt |
| Apply the KITTI-trained model directly, without Make3D tuning | Ours (Pose-CNN + DDVO) · Make3D test set | RMSE: 10.47 -> 8.09 | Helped |
Closest papers by meaning
- Unsupervised Learning of Monocular Depth Estimation and Visual Odometry With Deep Feature Reconstruction (CVPR 2018 · similarity 0.91)
- D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometry (CVPR 2020 · similarity 0.91)
- Camera Pose Matters: Improving Depth Prediction by Mitigating Pose Distribution Bias (CVPR 2021 · similarity 0.91)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.90)
- Depth Prediction without the Sensors: Leveraging Structure for Unsupervised Learning from Monocular Videos (AAAI 2019 · similarity 0.90)
- DME: Unveiling the Bias for Better Generalized Monocular Depth Estimation (AAAI 2024 · 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.