Research map / 3D vision / Face and shape
MonoArt: Progressive Structural Reasoning for Monocular Articulated 3D Reconstruction
Haitian Li, Haozhe Xie, Junxiang Xu, Beichen Wen, Fangzhou Hong, Ziwei Liu
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 Part-Aware Semantic Reasoner (ablation) | MonoArt · PartNet-Mobility (46 classes) | Type Acc. (%): 24.72 -> 67.47 | Improved |
| use triplet loss instead of cross-entropy in the reasoner (ablation) | MonoArt · PartNet-Mobility (46 classes) | Type Acc. (%): 57.74 -> 67.47 | Improved |
| add part-level supervision loss to the reasoner (vs none) (ablation) | MonoArt · PartNet-Mobility (46 classes) | Pivot Err.: 0.323 -> 0.108 | Improved |
| add dual-query initialization (ablation) | MonoArt · PartNet-Mobility (46 classes) | Type Acc. (%): 44.06 -> 67.47 | Improved |
| residual updates on both position and content queries vs one branch (ablation) | MonoArt · PartNet-Mobility (46 classes) | F-Score: 0.640 -> 0.670 | Improved |
| increase refinement layers from 0 to 6 (ablation) | MonoArt · PartNet-Mobility (46 classes) | Axis Err. (rad): 0.640 -> 0.423 | Improved |
| increase refinement layers from 6 to 9 (ablation) | MonoArt · PartNet-Mobility (46 classes) | Pivot Err.: 0.108 -> 0.161 | Hurt |
| predict joint origin as residual from position query centroid (ablation) | MonoArt · PartNet-Mobility (46 classes) | Pivot Err.: 0.191 -> 0.108 | Improved |
| include point embedding H in parameter regression (ablation) | MonoArt · PartNet-Mobility (46 classes) | Type Acc. (%): 27.14 -> 67.47 | Improved |
| MonoArt vs PhysX-Anything on joint axis error | MonoArt · PartNet-Mobility (46 classes) | Axis Err. (rad): 0.289 -> 0.423 | Hurt |
| MonoArt vs SINGAPO | MonoArt · PartNet-Mobility (7 classes) | F-Score: 0.572 -> 0.728 | Improved |
Closest papers by meaning
- Learning Monocular 3D Reconstruction of Articulated Categories From Motion (CVPR 2021 · similarity 0.90)
- Articulated Object Reconstruction from Rest-State Observation (ECCV 2026 · similarity 0.90)
- Articulation in Prime: Primitive-Based Articulated Object Understanding from a Single Casual Video (NeurIPS 2026 · similarity 0.89)
- UniCon3R: Unified Contact-aware 4D Human-Scene Reconstruction from Monocular Video (NeurIPS 2026 · similarity 0.88)
- S3O: A Dual-Phase Approach for Reconstructing Dynamic Shape and Skeleton of Articulated Objects from Single Monocular Video (ICML 2024 · similarity 0.88)
- SINGAPO: Single Image Controlled Generation of Articulated Parts in Objects (ICLR 2025 · 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.
