Research map / Detection and segmentation / Open-set and few-shot detection
SiM3D: Single-instance Multiview Multimodal and Multisetup 3D Anomaly Detection Benchmark
Alex Costanzino, Pierluigi Zama Ramirez, Luigi Lella, Matteo Ragaglia, Alessandro Oliva, Giuseppe Lisanti, Luigi Di Stefano
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 |
|---|---|---|---|
| use depth maps instead of point clouds as 3D input (ablation) | BTF · SiM3D real2real | I-AUROC: 0.478 -> 0.707 | Improved |
| use depth maps instead of point clouds as 3D input (ablation) | BTF · SiM3D real2real | V-AUPRO@1%: 0.444 -> 0.609 | Improved |
| use depth maps instead of point clouds as 3D input (ablation) | CFM · SiM3D real2real | I-AUROC: 0.448 -> 0.548 | Improved |
| use depth maps instead of point clouds as 3D input (ablation) | M3DM · SiM3D synth2real | I-AUROC: 0.402 -> 0.254 | Hurt |
| add depth to RGB input (ablation) | AST · SiM3D real2real | I-AUROC: 0.584 -> 0.687 | Improved |
| add depth to RGB input (ablation) | AST · SiM3D synth2real | I-AUROC: 0.544 -> 0.495 | Hurt |
| add point cloud features to RGB (multimodal) versus RGB only | PatchCore-family vs M3DM · SiM3D real2real | I-AUROC: 0.754 -> 0.621 | Hurt |
| train on synthetic renders instead of real scans | PatchCore (WRN-101) · SiM3D | I-AUROC: 0.754 -> 0.451 | Hurt |
| train on synthetic renders instead of real scans | AST (RGB+Depth) · SiM3D | V-AUPRO@1%: 0.699 -> 0.695 | No clear effect |
| use DINO-v2 instead of WRN-101 backbone (ablation) | PatchCore · SiM3D synth2real | I-AUROC: 0.451 -> 0.540 | Improved |
| use DINO-v2 instead of WRN-101 backbone (ablation) | PatchCore · SiM3D real2real | I-AUROC: 0.754 -> 0.678 | Hurt |
| use depth maps rather than point clouds as 3D input (ablation) | PatchCore · SiM3D real2real | I-AUROC: 0.415 -> 0.582 | Improved |
Closest papers by meaning
- Real-IAD D3: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detection (CVPR 2025 · similarity 0.90)
- Multimodal Industrial Anomaly Detection by Crossmodal Feature Mapping (CVPR 2024 · similarity 0.89)
- 3CAD: A Large-Scale Real-World 3C Product Dataset for Unsupervised Anomaly Detection (AAAI 2025 · similarity 0.88)
- Spotting the Unexpected (STU): A 3D LiDAR Dataset for Anomaly Segmentation in Autonomous Driving (CVPR 2025 · similarity 0.87)
- Looking 3D: Anomaly Detection with 2D-3D Alignment (CVPR 2024 · similarity 0.87)
- Towards Scalable 3D Anomaly Detection and Localization: A Benchmark via 3D Anomaly Synthesis and A Self-Supervised Learning Network (CVPR 2024 · 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.
