Research map / 3D vision / 3D generation
CoGeoAD: Hierarchical Color-Geometric Fusion with Multi-View Attention for Zero-Shot 3D Anomaly Detection
Ke Xu, Xinle Wang, Yanning Hou, Xueliang Ma, Juan Xie, Jianfeng Qiu
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 color branch (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 84.3 -> 87.4 | Improved |
| add geometric branch (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 83.6 -> 87.4 | Improved |
| add MS-CGF fusion (ablation) | CoGeoAD · MVTec3D-AD | AUPRO (%): 87.9 -> 91.9 | Improved |
| add MS-CGF fusion (ablation) | CoGeoAD · Eyecandies | I-AUROC (%): 82.3 -> 82.6 | Improved |
| add multi-view attention (vs mean pooling) (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 83.8 -> 87.4 | Improved |
| add multi-view attention (vs mean pooling) (ablation) | CoGeoAD · MVTec3D-AD | AUPRO (%): 88.2 -> 91.9 | Improved |
| add multi-view attention (ablation) | CoGeoAD · Eyecandies | I-AUROC (%): 78.9 -> 82.6 | Improved |
| use last layers 21-24 instead of shallow layers 1-12 (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 50.9 -> 87.4 | Improved |
| use last layers 21-24 instead of strided layers (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 86.8 -> 87.4 | Improved |
| use 9 views instead of 12 (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 82.1 -> 87.4 | Improved |
| use resolution 336 instead of 224 (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 75.1 -> 87.4 | Improved |
| add MVA attention vs MLP weighting (ablation) | CoGeoAD · MVTec3D-AD | I-AUROC (%): 85.0 -> 87.4 | Improved |
Closest papers by meaning
- Align3D-AD: Cross-Modal Feature Alignment and Dual-Prompt Learning for Zero-shot 3D Anomaly Detection (NeurIPS 2026 · similarity 0.84)
- Looking 3D: Anomaly Detection with 2D-3D Alignment (CVPR 2024 · similarity 0.84)
- PointAD: Comprehending 3D Anomalies from Points and Pixels for Zero-shot 3D Anomaly Detection (NeurIPS 2024 · similarity 0.84)
- Towards Generalizable 3D Anomaly Detection via Relational Inconsistency Modeling (NeurIPS 2026 · similarity 0.83)
- Multi-Sensor Object Anomaly Detection: Unifying Appearance, Geometry, and Internal Properties (CVPR 2025 · similarity 0.83)
- Real-IAD D3: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detection (CVPR 2025 · similarity 0.83)
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.
