Research map / 3D vision / Depth and stereo
The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth
Check 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 ManyDepth multi-frame cost-volume depth estimation instead of the strongest listed non-TTR baseline | KITTI Eigen split · KITTI Eigen split | Abs Rel: 0.102 -> 0.098 | Helped |
| use ManyDepth instead of the strongest listed Cityscapes baseline | Cityscapes test · Cityscapes test | Abs Rel: 0.119 -> 0.114 | Helped |
| add motion masking and consistency loss (ablation) | ManyDepth on KITTI · KITTI Eigen split | Abs Rel: 0.113 -> 0.098 | Helped |
| add the training augmentations alongside motion masking (ablation) | ManyDepth on KITTI · KITTI Eigen split | Abs Rel: 0.284 -> 0.113 | Helped |
| add the consistency teacher to masked training (ablation) | ManyDepth on KITTI · KITTI Eigen split | Abs Rel: 0.154 -> 0.098 | Helped |
| use a geometric cost volume rather than directly stacking two input frames (ablation) | ManyDepth on KITTI · KITTI Eigen split | Abs Rel: 0.121 -> 0.098 | Helped |
| use the proposed mask instead of the tested alternative masks (ablation) | ManyDepth on KITTI · KITTI Eigen split | Abs Rel: 0.099 -> 0.098 | Helped |
| add a third preceding frame to the cost volume (ablation) | ManyDepth on KITTI · KITTI Eigen split | Abs Rel: 0.098 -> 0.098 | No clear effect |
| add a future frame to the cost volume (ablation) | ManyDepth on KITTI · KITTI Eigen split | Abs Rel: 0.098 -> 0.097 | Helped |
| add static-camera and start-of-sequence augmentations (ablation) | ManyDepth in standard two-frame evaluation · KITTI Eigen split, standard input | Abs Rel: 0.100 -> 0.098 | Helped |
| add static-camera and start-of-sequence augmentations (ablation) | ManyDepth with single-frame input · KITTI Eigen split, start-of-sequence input | Abs Rel: 0.148 -> 0.118 | Helped |
| add static-camera and start-of-sequence augmentations (ablation) | ManyDepth with duplicate-frame input · KITTI Eigen split, static-camera input | Abs Rel: 0.158 -> 0.117 | Helped |
Closest papers by meaning
- Self-Supervised Monocular Trained Depth Estimation Using Self-Attention and Discrete Disparity Volume (CVPR 2020 · similarity 0.93)
- Multi-Frame Self-Supervised Depth With Transformers (CVPR 2022 · similarity 0.92)
- Mining Supervision for Dynamic Regions in Self-Supervised Monocular Depth Estimation (CVPR 2024 · similarity 0.91)
- SQLdepth: Generalizable Self-Supervised Fine-Structured Monocular Depth Estimation (AAAI 2024 · similarity 0.91)
- GeoDepth: From Point-to-Depth to Plane-to-Depth Modeling for Self-Supervised Monocular Depth Estimation (CVPR 2025 · similarity 0.90)
- From-Ground-To-Objects: Coarse-to-Fine Self-supervised Monocular Depth Estimation of Dynamic Objects with Ground Contact Prior (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.