Research map / 3D vision / Face and shape
PixARMesh: Autoregressive Mesh-Native Single-View Scene Reconstruction
Xiang Zhang, Sohyun Yoo, Hongrui Wu, Chuan Li, Jianwen Xie, Zhuowen Tu
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 |
|---|---|---|---|
| joint pose-mesh modeling instead of two-stage (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (object-level) | Chamfer Distance: CD 4.75 -> 4.04, F-Score 80.85 -> 82.27 | Improved |
| joint pose-mesh modeling instead of two-stage (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (scene-level) | F-Score: 33.32 -> 33.55 | Improved |
| predict layout in-sequence instead of post-hoc optimization (EdgeRunner-FT) (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (scene-level) | F-Score: 27.81 -> 33.55 | Improved |
| add pixel-aligned image features (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (scene-level) | Chamfer Distance: 57.78 -> 39.30 | Improved |
| add pixel-aligned image features (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (object-level) | F-Score: 77.22 -> 84.64 | Improved |
| add scene context aggregation on top of image features (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (scene-level) | F-Score: 44.67 -> 46.15 | Improved |
| add scene context aggregation on top of image features (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (object-level) | F-Score: 84.64 -> 82.27 | Hurt |
| add scene context aggregation on top of image features (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (scene-level) | CD-S: 16.12 -> 18.52 | Hurt |
| add scene context aggregation on top of image features (ablation) | PixARMesh-BPT · 3D-FRONT (scene-level) | Chamfer Distance: 57.03 -> 49.82 | Improved |
| add pixel-aligned image features (ablation) | PixARMesh-BPT · 3D-FRONT (scene-level) | F-Score: 37.98 -> 42.18 | Improved |
| add scene context aggregation without image features (ablation) | PixARMesh-BPT · 3D-FRONT (object-level) | Chamfer Distance: 5.34 -> 5.85 | Hurt |
| use ground-truth depth instead of predicted (ablation) | PixARMesh-EdgeRunner · 3D-FRONT (object-level) | F-Score: 81.64 -> 86.66 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- DepR — DepR: Depth Guided Single-view Scene Reconstruction with Instance-level Diffusion (ICCV 2025) · better on 4 of 6 shared results
- MIDI — MIDI: Multi-Instance Diffusion for Single Image to 3D Scene Generation (CVPR 2025) · better on 4 of 4 shared results
- Uni-3D — Uni-3D: A Universal Model for Panoptic 3D Scene Reconstruction (ICCV 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ARMesh: Autoregressive Mesh Generation via Next-Level-of-Detail Prediction (NeurIPS 2025 · similarity 0.87)
- Coherent 3D Scene Diffusion From a Single RGB Image (NeurIPS 2024 · similarity 0.86)
- PivotMesh: Generic 3D Mesh Generation via Pivot Vertices Guidance (ICLR 2025 · similarity 0.86)
- LiteReality: Graphic-Ready 3D Scene Reconstruction from RGB-D Scans (NeurIPS 2025 · similarity 0.86)
- AutoPartGen: Autoregressive 3D Part Generation and Discovery (NeurIPS 2025 · similarity 0.86)
- MeshAnything: Artist-Created Mesh Generation with Autoregressive Transformers (ICLR 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.
