Research map / Point clouds / Point cloud classification
R3D-AD: Reconstruction via Diffusion for 3D Anomaly Detection
Zheyuan Zhou, Le Wang, Naiyu Fang, Zili Wang, Lemiao Qiu, Shuyou Zhang
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 latent shape embedding condition to vanilla DDPM (ablation) | R3D-AD · Real3D-AD | I-AUROC: 0.586 -> 0.667 | Improved |
| add latent shape embedding condition to vanilla DDPM (ablation) | R3D-AD · Real3D-AD | P-AUROC: 0.524 -> 0.513 | No clear effect |
| predict relative point displacements instead of absolute shape (ablation) | R3D-AD · Real3D-AD | P-AUROC: 0.513 -> 0.573 | Improved |
| predict relative point displacements instead of absolute shape (ablation) | R3D-AD · Real3D-AD | I-AUROC: 0.667 -> 0.712 | Improved |
| add Patch-Gen anomaly simulation (ablation) | R3D-AD · Real3D-AD | I-AUROC: 0.712 -> 0.734 | Improved |
| Patch-Gen selection ratio raised from 1/32 to 1/8 (ablation) | Patch-Gen · Real3D-AD | I-AUROC: 0.734 -> 0.683 | Hurt |
| Patch-Gen selection ratio lowered from 1/32 to 1/64 (ablation) | Patch-Gen · Real3D-AD | I-AUROC: 0.734 -> 0.716 | Hurt |
| increase Patch-Gen scaling factor from 0.1 to 0.8 (ablation) | Patch-Gen · Real3D-AD | I-AUROC: 0.734 -> 0.661 | Hurt |
| replace memory-bank/embedding methods with one-step diffusion reconstruction on raw points | 3D anomaly detection · Anomaly-ShapeNet | I-AUROC (average): 0.659 -> 0.749 | Improved |
| replace memory-bank/embedding methods with one-step diffusion reconstruction on raw points | 3D anomaly detection · Real3D-AD | I-AUROC (average): 0.725 -> 0.734 | Improved |
| train on a subset of bowl categories and test on unseen bowl4 | R3D-AD · Anomaly-ShapeNet | I-AUROC: 0.744 -> 0.715 | Mixed |
| train on ShapeNetCore.v2 and test on Real3D-AD (cross-dataset) | R3D-AD · Real3D-AD Airplane | I-AUROC: 0.772 -> 0.614 | Hurt |
Later papers that compared with this method
4 later accepted papers report R3D-AD as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Anomaly Factory 3D: A Modular Framework for Diverse Pseudo-Anomaly Synthesis in Unsupervised 3D Anomaly Detection (ECCV 2026 · better on 2 of 2 shared results)
- Back to Point: Exploring Point-Language Models for Zero-Shot 3D Anomaly Detection (CVPR 2026 · better on 0 of 1 shared result)
- PO3AD: Predicting Point Offsets toward Better 3D Point Cloud Anomaly Detection (CVPR 2025 · better on 1 of 1 shared result)
- Bridging 3D Anomaly Localization and Repair via High-Quality Continuous Geometric Representation (ICCV 2025 · better on 1 of 1 shared result)
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
- Towards Scalable 3D Anomaly Detection and Localization: A Benchmark via 3D Anomaly Synthesis and A Self-Supervised Learning Network (CVPR 2024 · similarity 0.90)
- Registration is a Powerful Rotation-Invariance Learner for 3D Anomaly Detection (NeurIPS 2025 · similarity 0.88)
- AnomalyDiffusion: Few-Shot Anomaly Image Generation with Diffusion Model (AAAI 2024 · similarity 0.88)
- Towards Generalizable 3D Anomaly Detection via Relational Inconsistency Modeling (NeurIPS 2026 · similarity 0.88)
- PO3AD: Predicting Point Offsets toward Better 3D Point Cloud Anomaly Detection (CVPR 2025 · similarity 0.87)
- Real-IAD D3: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detection (CVPR 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.
