Research map / Video / Video segmentation and tracking
FutureDepth: Learning to Predict the Future Improves Video Depth Estimation
Rajeev Yasarla, Manish Kumar Singh, Hong Cai, Yunxiao Shi, Jisoo Jeong, Yinhao Zhu, Shizhong Han, Risheek Garrepalli, Fatih Porikli
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 |
|---|---|---|---|
| naively extend single-frame model to multi-frame batch processing (ablation) | baseline Swin-L · KITTI | OPW: 0.544 -> 0.540 | No clear effect |
| add F-Net future prediction (ablation) | multi-frame baseline · KITTI | OPW: 0.540 -> 0.311; Sq Rel 0.154 -> 0.129 | Improved |
| add R-Net with adaptive masking instead of random masking (ablation) | multi-frame baseline · KITTI | OPW: 0.478 -> 0.416 | Improved |
| use adaptive masking instead of random or tube masking in R-Net (ablation) | R-Net · KITTI | Abs Rel: 0.051 -> 0.048 | Improved |
| combine F-Net and R-Net (ablation) | multi-frame baseline · KITTI | OPW: 0.311 -> 0.284 | Improved |
| add progressive refinement network (ablation) | FutureDepth · KITTI | Sq Rel: 0.122 -> 0.119 | Improved |
| increase future prediction steps L from 0 to 4 (ablation) | F-Net · KITTI | RMSE: 1.999 -> 1.931 | Improved |
| increase number of frames T from 6 to 8 (ablation) | FutureDepth · KITTI | RMSE: 1.906 -> 1.911 | No clear effect |
| use very high inference masking ratio 0.9 in adaptive sampler (ablation) | R-Net · KITTI | RMSE: 1.920 -> 1.956 | Hurt |
| increase refinement steps N from 0 to 3 (ablation) | FutureDepth · KITTI | RMSE: 1.931 -> 1.920 | Improved |
Closest papers by meaning
- ICDepth: Taming Video Diffusion Models for Video Depth Estimation via In-Context Conditioning (ECCV 2026 · similarity 0.88)
- The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth (CVPR 2021 · similarity 0.87)
- FlashDepth: Real-time Streaming Video Depth Estimation at 2K Resolution (ICCV 2025 · similarity 0.87)
- GemDepth: Geometry-Embedded Features for 3D-Consistent Video Depth (ICML 2026 · similarity 0.87)
- Video Depth Anything: Consistent Depth Estimation for Super-Long Videos (CVPR 2025 · similarity 0.86)
- Learning Temporally Consistent Video Depth from Video Diffusion Priors (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.
