Research map / 3D vision / Depth and stereo
HFD-Teacher: High-Frequency Depth Distillation from Depth Foundation Models for Enhanced Depth Completion
Zhiyuan Yang, Anqi Cheng, Haiyue Zhu, Tianjiao Li, Pey Yuen Tao, Kezhi Mao
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 fixed level-3 wavelet decomposition to baseline (ablation) | Baseline ViT-S · NYUD-v2 ablation | eps_acc: 1.622 -> 1.501 | Improved |
| use fixed level 5 instead of 3 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | process time: 0.059 -> 0.090 | Hurt |
| use dynamic levels 1,2,3 instead of fixed level 3 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_acc: 1.501 -> 1.385 | Improved |
| use dynamic levels 2,3 instead of 1,2,3 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_acc: 1.385 -> 1.445 | Hurt |
| add local partitioning (14x14 windows) (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_acc: 1.385 -> 1.285 | Improved |
| enlarge window from 14x14 to 24x24 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_acc: 1.285 -> 1.330 | Hurt |
| shrink window from 14x14 to 8x8 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_acc: 1.285 -> 1.300 | Hurt |
| add topological constraint theta=0.1 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_acc: 1.285 -> 1.014 | Improved |
| raise topological threshold theta from 0.1 to 0.2 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_acc: 1.014 -> 1.280 | Hurt |
| add topological constraint theta=0.1 (ablation) | HFD-Teacher (ViT-S) · NYUD-v2 ablation | eps_comp: 1.805 -> 1.457 | Improved |
| scale backbone from ViT-S to ViT-L | HFD-Teacher · NYUD-v2 | RMSE: 0.107 -> 0.087 | Improved |
| scale backbone from ViT-S to ViT-L | HFD-Teacher · KITTI-DC test | RMSE (mm): 710.51 -> 679.98 | Improved |
Closest papers by meaning
- Distilling Monocular Foundation Model for Fine-grained Depth Completion (CVPR 2025 · similarity 0.88)
- Monitored Distillation for Positive Congruent Depth Completion (ECCV 2022 · similarity 0.87)
- Zero-shot Depth Completion via Test-time Alignment with Affine-invariant Depth Prior (AAAI 2025 · similarity 0.86)
- Deep Depth Completion of a Single RGB-D Image (CVPR 2018 · similarity 0.86)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.86)
- SVDC: Consistent Direct Time-of-Flight Video Depth Completion with Frequency Selective Fusion (CVPR 2025 · similarity 0.85)
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.
