Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Test-Time Prompt Tuning for Zero-Shot Depth Completion
Chanhwi Jeong, Inhwan Bae, Jin-Hwi Park, Hae-Gon Jeon
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 |
|---|---|---|---|
| replace least squares scale fitting with test-time fine-tuning on sparse depth | UniDepthV2 · IBims-1 | MAE: 0.113 -> 0.049 | Improved |
| replace fine-tuning with pixelwise visual prompt | UniDepthV2 · DDAD | MAE: 1.775 -> 1.482 | Improved |
| replace fine-tuning with pixelwise visual prompt | UniDepthV2 · IBims-1 | RMSE: 0.152 -> 0.161 | Hurt |
| replace fine-tuning with pixelwise visual prompt | UniDepthV2 · KITTI | RMSE: 1.329 -> 1.333 | No clear effect |
| use pixelwise prompt instead of random patch prompt (ablation) | UniDepthV2 · IBims-1 | MAE: 0.111 -> 0.043 | Improved |
| use pixelwise prompt instead of fixed patch prompt (ablation) | UniDepthV2 · IBims-1 | MAE: 0.079 -> 0.043 | Improved |
| use pixelwise prompt instead of padding prompt (ablation) | UniDepthV2 · IBims-1 | MAE: 0.050 -> 0.043 | Improved |
| use pixelwise prompt instead of full fine-tuning (learnable parameters) (ablation) | UniDepthV2 · IBims-1 | learnable parameters: 100% -> 0.25% | Improved |
| add visual prompt tuning on top of least squares fit (ablation) | ZoeDepth · IBims-1 | MAE: 0.623 -> 0.176 | Improved |
| add visual prompt tuning on top of least squares fit (ablation) | Metric3dV2 · NYUv2 | MAE: 0.174 -> 0.069 | Improved |
| use visual prompt rather than Marigold-DC under very sparse input (5 samples) | Ours-VP · NYUv2 (5 samples) | MAE: 0.358 -> 0.138 | Improved |
| use visual prompt rather than Marigold-DC under 4 LiDAR channels | Ours-VP · KITTI (4 channels) | MAE: 1.669 -> 0.812 | Improved |
Closest papers by meaning
- Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Prior (AAAI 2025 · similarity 0.92)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.89)
- Depth Prompting for Sensor-Agnostic Depth Estimation (CVPR 2024 · similarity 0.89)
- Distilling Monocular Foundation Model for Fine-grained Depth Completion (CVPR 2025 · similarity 0.89)
- Prompting Depth Anything for 4K Resolution Accurate Metric Depth Estimation (CVPR 2025 · similarity 0.88)
- Depth Pro: Sharp Monocular Metric Depth in Less Than a Second (ICLR 2025 · 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.
