Research map / Video / Video segmentation and tracking
Exploiting Temporal Consistency for Real-Time Video Depth Estimation
Haokui Zhang, Chunhua Shen, Ying Li, Yuanzhouhan Cao, Yu Liu, Youliang Yan
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 |
|---|---|---|---|
| add CLSTM temporal modelling with 3 input frames instead of a 2D CNN (ablation) | 2DCNN · NYU Depth V2 | Rel: 0.139 -> 0.134 | Improved |
| increase input frames from 3 to 5 (ablation) | ST-CLSTM · NYU Depth V2 | Rel: 0.134 -> 0.132 | Improved |
| increase input frames from 3 to 5 (ablation) | ST-CLSTM · NYU Depth V2 | delta1: 0.824 -> 0.833 | Improved |
| replace the 3-layer conv 2DCNN with ST-CLSTM (5 frames) (ablation) | 2DCNN · NYU Depth V2 | RMS: 0.585 -> 0.572 | Improved |
| replace the 2DCNN with ST-CLSTM (5 frames) (ablation) | 2DCNN · KITTI | Rel: 0.111 -> 0.104 | Improved |
| replace the 2DCNN with ST-CLSTM (5 frames) (ablation) | 2DCNN · KITTI | RMS: 4.385 -> 4.139 | Improved |
| add 3D CNN generative adversarial temporal loss (ablation) | ST-CLSTM · NYU Depth V2 | Rel: 0.132 -> 0.131 | No clear effect |
| add 3D CNN generative adversarial temporal loss (ablation) | ST-CLSTM · KITTI | Rel: 0.104 -> 0.101 | Improved |
| add 3D CNN generative adversarial temporal loss (ablation) | ST-CLSTM · KITTI | delta1: 0.883 -> 0.890 | Improved |
| add ST-CLSTM temporal modelling (ablation) | 2DCNN baseline · NYU Depth V2 | TCC: 0.846 -> 0.866 | Improved |
| add 3D CNN discriminator temporal loss (ablation) | ST-CLSTM · NYU Depth V2 | TCC: 0.866 -> 0.870 | Improved |
| add 3D CNN discriminator temporal loss (ablation) | ST-CLSTM · NYU Depth V2 | TMC: 0.962 -> 0.965 | Improved |
Closest papers by meaning
- Video Depth Anything: Consistent Depth Estimation for Super-Long Videos (CVPR 2025 · similarity 0.89)
- Learning Temporally Consistent Video Depth from Video Diffusion Priors (CVPR 2025 · similarity 0.88)
- MAMo: Leveraging Memory and Attention for Monocular Video Depth Estimation (ICCV 2023 · similarity 0.88)
- Robust Consistent Video Depth Estimation (CVPR 2021 · similarity 0.88)
- Buffer Anytime: Zero-Shot Video Depth and Normal from Image Priors (CVPR 2025 · similarity 0.87)
- Align3R: Aligned Monocular Depth Estimation for Dynamic Videos (CVPR 2025 · similarity 0.87)
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.
