Research map / 3D vision / NeRF and radiance fields
SlotLifter: Slot-guided Feature Lifting for Learning Object-Centric Radiance Fields
Yu Liu, Baoxiong Jia, Yixin Chen, Siyuan Huang
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 slot-guided feature lifting (lifted 2D features as point features) (ablation) | SlotLifter · ScanNet | PSNR: 11.60 -> 28.36 | Improved |
| add feature lifting (ablation) | SlotLifter · Room-Diverse | FG-ARI: 65.3 -> 84.8 | Improved |
| add random masking of lifted features (ablation) | SlotLifter · Room-Diverse | FG-ARI: 43.6 -> 84.8 | Improved |
| add random masking of lifted features (ablation) | SlotLifter · Room-Diverse | PSNR: 29.89 -> 29.80 | Hurt |
| add slot-based density instead of MLP density (ablation) | SlotLifter · ScanNet | NV-FG-ARI: 6.6 -> 31.1 | Improved |
| lengthen random masking decay from 0 to 30000 steps (ablation) | SlotLifter · Room-Diverse | ARI: 85.8 -> 90.3 | Improved |
| lengthen masking decay from 30000 to 100000 steps (ablation) | SlotLifter · Room-Diverse | PSNR: 29.80 -> 28.68 | Hurt |
| image-based feature lifting versus slot-only decoders (BO-uORF, OSRT) on real scenes | 3D object-centric models · ScanNet | PSNR: 13.34 (OSRT) -> 28.36 | Improved |
| SlotLifter versus sVORF on CLEVR-567 rendering | novel-view synthesis · CLEVR-567 | PSNR: 37.20 -> 36.09 | Hurt |
Closest papers by meaning
- Slot-guided Volumetric Object Radiance Fields (NeurIPS 2023 · similarity 0.92)
- Shepherding Slots to Objects: Towards Stable and Robust Object-Centric Learning (CVPR 2023 · similarity 0.88)
- GRF: Learning a General Radiance Field for 3D Representation and Rendering (ICCV 2021 · similarity 0.87)
- GLASS: Guided Latent Slot Diffusion for Object-Centric Learning (CVPR 2025 · similarity 0.86)
- DynaVol: Unsupervised Learning for Dynamic Scenes through Object-Centric Voxelization (ICLR 2024 · similarity 0.86)
- Rethinking End-to-End 2D to 3D Scene Segmentation in Gaussian Splatting (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.
