Research map / 3D vision / Depth and stereo
Transfusion: A Novel SLAM Method Focused on Transparent Objects
Yifan Zhu, Jiaxiong Qiu, Bo Ren
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 |
|---|---|---|---|
| cut transparent objects from depth input with TC-ICP (vs ElasticFusion) | ElasticFusion · Trans-SLAM (mean) | ATE: 0.560 -> 0.219 | Improved |
| cut transparent objects from depth input with TC-ICP (vs ElasticFusion) | ElasticFusion · Trans-SLAM (mean) | R.RPE: 0.109 -> 0.027 | Improved |
| cut transparent objects from depth input with TC-ICP (vs ElasticFusion) | ElasticFusion · Trans-SLAM (mean) | T.RPE: 0.220 -> 0.080 | Improved |
| add mask adjustment (ablation) | Transfusion · Trans-SLAM (mean) | ATE: 0.345 -> 0.301 | Improved |
| add mask adjustment (ablation) | Transfusion · Trans-SLAM (mean) | R.RPE: 0.164 -> 0.034 | Improved |
| add mask adjustment (ablation) | Transfusion · Trans-SLAM (mean) | T.RPE: 0.146 -> 0.110 | Improved |
| add mask adjustment (ablation) | Transfusion · Trans-SLAM Cylindrical Glass | R.RPE: 0.056 -> 0.084 | Hurt |
| use the proposed two-branch segmentation network instead of SINet (ablation) | Transfusion · Trans-SLAM (mean) | ATE: 0.468 -> 0.301 | Improved |
| use the proposed two-branch segmentation network instead of SINet (ablation) | Transfusion · Trans-SLAM (mean) | R.RPE: 0.086 -> 0.034 | Improved |
| use the proposed two-branch segmentation network instead of SINet (ablation) | Transfusion · Trans-SLAM (mean) | T.RPE: 0.213 -> 0.110 | Improved |
| use SINet segmentation network instead of no mask adjustment (ablation) | Transfusion · Trans-SLAM (mean) | ATE: 0.345 -> 0.468 | Hurt |
| use Transfusion on scenes without transparent objects | ElasticFusion · TUM fr1/desk | ATE: 0.020 -> 0.021 | No clear effect |
Closest papers by meaning
- VGGT-SLAM: Dense RGB SLAM Optimized on the SL(4) Manifold (NeurIPS 2025 · similarity 0.86)
- SLAM3R: Real-Time Dense Scene Reconstruction from Monocular RGB Videos (CVPR 2025 · similarity 0.85)
- GS-SLAM: Dense Visual SLAM with 3D Gaussian Splatting (CVPR 2024 · similarity 0.85)
- Gaussian Splatting SLAM (CVPR 2024 · similarity 0.85)
- VTGaussian-SLAM: RGBD SLAM for Large Scale Scenes with Splatting View-Tied 3D Gaussians (ICML 2025 · similarity 0.84)
- IBD-SLAM: Learning Image-Based Depth Fusion for Generalizable SLAM (CVPR 2024 · similarity 0.84)
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.
