Research map / 3D vision / Depth and stereo
TransformerFusion: Monocular RGB Scene Reconstruction using Transformers
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use transformer-based learned view fusion rather than average fusion (ablation) | TransformerFusion · ScanNet test | F-score: 0.557 -> 0.655 | Helped |
| Use transformer-based learned view fusion rather than MLP-predicted weighted averaging (ablation) | TransformerFusion · ScanNet test | F-score: 0.583 -> 0.655 | Helped |
| Add spatial feature refinement (ablation) | TransformerFusion · ScanNet test | F-score: 0.361 -> 0.655 | Helped |
| Use attention-based selection instead of random selection for 4 views (ablation) | TransformerFusion · ScanNet test | F-score: 0.502 -> 0.581 | Helped |
| Use attention-based selection instead of random selection for 8 views (ablation) | TransformerFusion · ScanNet test | F-score: 0.596 -> 0.636 | Helped |
| Use attention-based selection instead of random selection for 16 views (ablation) | TransformerFusion · ScanNet test | F-score: 0.638 -> 0.655 | Helped |
| Use 16 selected views rather than 8 (ablation) | TransformerFusion · ScanNet test | F-score: 0.636 -> 0.655 | Helped |
| Use 16 selected views rather than 4 (ablation) | TransformerFusion · ScanNet test | F-score: 0.581 -> 0.655 | Helped |
| Add coarse-to-fine filtering (ablation) | TransformerFusion · ScanNet test | F-score: 0.631 -> 0.655 | Helped |
| Include projected depth in transformer inputs (ablation) | TransformerFusion · ScanNet test | F-score: 0.525 -> 0.655 | Helped |
| Include the viewing ray in transformer inputs (ablation) | TransformerFusion · ScanNet test | F-score: 0.621 -> 0.655 | Helped |
| Use TransformerFusion rather than NeuralRecon | 3D scene reconstruction · ScanNet test | Chamfer: 7.11 -> 6.89 | Helped |
Closest papers by meaning
- Coherent 3D Scene Diffusion From a Single RGB Image (NeurIPS 2024 · similarity 0.87)
- SLAM3R: Real-Time Dense Scene Reconstruction from Monocular RGB Videos (CVPR 2025 · similarity 0.87)
- O^2-Recon: Completing 3D Reconstruction of Occluded Objects in the Scene with a Pre-trained 2D Diffusion Model (AAAI 2024 · similarity 0.87)
- VGGT: Visual Geometry Grounded Transformer (CVPR 2025 · similarity 0.87)
- 4DGT: Learning a 4D Gaussian Transformer Using Real-World Monocular Videos (NeurIPS 2025 · similarity 0.87)
- MorpheuS: Neural Dynamic 360deg Surface Reconstruction from Monocular RGB-D Video (CVPR 2024 · 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.