Research map / 3D vision / Depth and stereo
Depth From Videos in the Wild: Unsupervised Monocular Depth Learning From Unknown Cameras
Ariel Gordon, Hanhan Li, Rico Jonschkowski, Anelia Angelova
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 |
|---|---|---|---|
| learn camera intrinsics instead of using given ones | depth model · KITTI Eigen | Abs Rel: 0.129 -> 0.128 | Improved |
| learn camera intrinsics instead of using given ones | depth model · Cityscapes | Abs Rel: 0.129 -> 0.127 | Improved |
| learn intrinsics, odometry without correction | odometry model · KITTI odometry Seq. 09 | t_rel: 3.1% -> 7.5% | Hurt |
| add inference-time correction using known focal length to learned intrinsics | odometry model · KITTI odometry Seq. 09 | t_rel: 7.5% -> 2.7% | Improved |
| pool Cityscapes and KITTI for training with learned intrinsics | depth model · Cityscapes | Abs Rel: 0.127 -> 0.121 | Improved |
| add object motion modeling (ablation) | depth model · Cityscapes (trained on Cityscapes+KITTI) | Abs Rel: 0.172 -> 0.121 | Improved |
| add object motion modeling (ablation) | depth model · KITTI Eigen (trained on Cityscapes+KITTI) | Abs Rel: 0.130 -> 0.124 | Improved |
| add occlusion-aware loss (ablation) | depth model · Cityscapes (trained on Cityscapes+KITTI) | Abs Rel: 0.127 -> 0.121 | Improved |
| add occlusion-aware loss (ablation) | depth model · KITTI Eigen (trained on Cityscapes+KITTI) | Abs Rel: 0.126 -> 0.124 | Improved |
| add randomized layer normalization (ablation) | depth model · Cityscapes (trained on Cityscapes+KITTI) | Abs Rel: 0.124 -> 0.121 | Improved |
| use segmentation masks instead of bounding boxes for possibly-mobile mask (ablation) | depth model · Cityscapes (trained on Cityscapes+KITTI) | Abs Rel: 0.120 -> 0.121 | No clear effect |
Closest papers by meaning
- Learning Depth From Monocular Videos Using Direct Methods (CVPR 2018 · similarity 0.93)
- Depth Prediction without the Sensors: Leveraging Structure for Unsupervised Learning from Monocular Videos (AAAI 2019 · similarity 0.92)
- Seurat: From Moving Points to Depth (CVPR 2025 · similarity 0.89)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.89)
- Video Depth Anything: Consistent Depth Estimation for Super-Long Videos (CVPR 2025 · similarity 0.89)
- Mining Supervision for Dynamic Regions in Self-Supervised Monocular Depth Estimation (CVPR 2024 · similarity 0.89)
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.
