Research map / 3D vision / 3D generation
3D-IDE: 3D Implicit Depth Emergent
Chushan Zhang, Ruihan Lu, Jinguang Tong, Yikai Wang, Hongdong Li
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 global scene-level alignment loss to a frozen 3D teacher (ablation) | 3D-IDE · ScanRefer | Acc@0.25: 53.7 -> 56.9 | Improved |
| add global scene-level alignment loss (ablation) | 3D-IDE · Multi3DRefer | F1@0.25: 46.0 -> 55.6 | Improved |
| add geometric depth-validator loss on top of global loss (ablation) | 3D-IDE · ScanRefer | Acc@0.25: 56.9 -> 59.8 | Improved |
| add geometric depth-validator loss on top of global loss (ablation) | 3D-IDE · Multi3DRefer | F1@0.25: 55.6 -> 59.7 | Improved |
| use from-scratch validator instead of pretrained VGGT depth head (ablation) | 3D-IDE · ScanRefer | Acc@0.25: 59.6 -> 59.8 | No clear effect |
| use from-scratch validator instead of pretrained VGGT depth head (ablation) | 3D-IDE · Multi3DRefer | F1@0.5: 53.2 -> 54.3 | No clear effect |
| add localized cross-view consistency loss (ablation) | 3D-IDE · ScanRefer | Acc@0.25: 59.8 -> 60.9 | Improved |
| add localized cross-view consistency loss (ablation) | 3D-IDE · Multi3DRefer | F1@0.5: 54.3 -> 54.9 | Improved |
| add IGEP to RGB-only baseline (ablation) | 3D-IDE · NAVI spatial correspondence | Recall@2cm: 40.15 -> 42.27 | Improved |
| add IGEP to RGB-only baseline (ablation) | 3D-IDE · NAVI normal prediction | RMSE: 32.26 -> 31.36 | Improved |
| withhold 3D inputs at inference from Video-3D LLM | 3D-IDE · Scan2Cap | C@0.5: 83.8 -> 31.5 | Hurt |
| use 3D-IDE vs VG LLM-8B for inference cost | 3D-IDE · Inference | mean time (s): 3.60 -> 1.61 | Improved |
Closest papers by meaning
- Imagine3D-LLM: Teaching MLLMs to Imagine 3D Scenes Before Answering (NeurIPS 2026 · similarity 0.90)
- Inst3D-LMM: Instance-Aware 3D Scene Understanding with Multi-modal Instruction Tuning (CVPR 2025 · similarity 0.90)
- Surprise3D: A Dataset for Spatial Understanding and Reasoning in Complex 3D Scenes (NeurIPS 2025 · similarity 0.89)
- Think with 3D: Geometric Imagination Grounded Spatial Reasoning from Limited Views (CVPR 2026 · similarity 0.89)
- Probing the 3D Awareness of Visual Foundation Models (CVPR 2024 · similarity 0.89)
- Beyond 3D VQAs: Injecting 3D Spatial Priors into Vision-Language Models for Enhanced Geometric Reasoning (CVPR 2026 · similarity 0.89)
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.
