Research map / Label-efficient and robust learning / Knowledge distillation
UNIKD: UNcertainty-Filtered Incremental Knowledge Distillation for Neural Implicit Representation
Mengqi Guo, Chen Li, Hanlin Chen, Gim Hee Lee
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 student-teacher knowledge distillation to incremental MonoSDF (ablation) | MonoSDF · ICL-NUIM | F1: 64.71 -> 90.32 | Improved |
| Add uncertainty-based filter to teacher queries (ablation) | UNIKD · ICL-NUIM | F1: 81.63 -> 90.32 | Improved |
| Add uncertainty-based filter to teacher queries (ablation) | UNIKD with NeRF · 360Capture | PSNR: 19.39 -> 21.21 | Improved |
| Store camera poses of previous time steps in addition (ablation) | UNIKD · ICL-NUIM | F1: 90.32 vs 89.50 | No clear effect |
| Apply UNIKD to Tri-MipRF backbone | Tri-MipRF · ScanNet | PSNR: 20.34 -> 27.02 | Improved |
| Apply UNIKD to ZipNeRF backbone | ZipNeRF · ScanNet | PSNR: 21.78 -> 27.85 | Improved |
| Use UNIKD instead of replay-based KR (no stored data) | incremental MonoSDF · Replica | F1: 79.67 -> 86.52; memory 30 MB -> 3 MB | Improved |
| UNIKD (MonoSDF) versus NICE-SLAM on Replica Chamfer Distance | incremental 3D reconstruction · Replica | Chamfer Distance: 3.11 vs 2.93 | Hurt |
| UNIKD versus keyframe replay KR on 360Capture D6 PSNR | incremental NeRF · 360Capture D6 | PSNR: 21.07 vs 21.14 | Hurt |
Closest papers by meaning
- Putting NeRF on a Diet: Semantically Consistent Few-Shot View Synthesis (ICCV 2021 · similarity 0.88)
- UMAMI: Unifying Masked Autoregressive Models and Deterministic Rendering for View Synthesis (NeurIPS 2025 · similarity 0.87)
- Neural Implicit Dictionary Learning via Mixture-of-Expert Training (ICML 2022 · similarity 0.87)
- 3D Reconstruction with Generalizable Neural Fields using Scene Priors (ICLR 2024 · similarity 0.87)
- Neural Experts: Mixture of Experts for Implicit Neural Representations (NeurIPS 2024 · similarity 0.86)
- GoLF-NRT: Integrating Global Context and Local Geometry for Few-Shot View Synthesis (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.
