# When Point Clouds Outperform Pixels: Rethinking Zero-Shot Multimodal Anomaly Detection

arXiv: https://arxiv.org/abs/2609.25793 · HTML: https://mlsanity.com/paper/2609.25793/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 18 of its 33 references that are in the map.
- First posted on arXiv 22 Sep 2026.

## Builds on (most cited first, 18 of 18)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation](https://mlsanity.com/paper/1606.04797/) (2016)
- [CLIP-Adapter: Better Vision-Language Models with Feature Adapters](https://mlsanity.com/paper/2110.04544/) (2021)
- [ULIP: Learning a Unified Representation of Language, Images, and Point Clouds for 3D Understanding](https://mlsanity.com/paper/2212.05171/) (2022)
- [CLIP-Driven Universal Model for Organ Segmentation and Tumor Detection](https://mlsanity.com/paper/2301.00785/) (2023)
- [ULIP-2: Towards Scalable Multimodal Pre-training for 3D Understanding](https://mlsanity.com/paper/2305.08275/) (2023)
- [Back to the Feature: Classical 3D Features are (Almost) All You Need for 3D Anomaly Detection](https://mlsanity.com/paper/2203.05550/) (2022)
- [Cheating Depth: Enhancing 3D Surface Anomaly Detection via Depth Simulation](https://mlsanity.com/paper/2311.01117/) (2023)
- [PointAD: Comprehending 3D Anomalies from Points and Pixels for Zero-shot 3D Anomaly Detection](https://mlsanity.com/paper/2410.00320/) (2024)
- [M3DM-NR: RGB-3D Noisy-Resistant Industrial Anomaly Detection via Multimodal Denoising](https://mlsanity.com/paper/2406.02263/) (2024)
- [DecAlign: Hierarchical Cross-Modal Alignment for Decoupled Multimodal Representation Learning](https://mlsanity.com/paper/2503.11892/) (2025)
- [Learning Feature Inversion for Multi-class Anomaly Detection under General-purpose COCO-AD Benchmark](https://mlsanity.com/paper/2404.10760/) (2024)
- [VisualAD: Language-Free Zero-Shot Anomaly Detection via Vision Transformer](https://mlsanity.com/paper/2603.07952/) (2026)
- [AnomalyMoE: Towards a Language-free Generalist Model for Unified Visual Anomaly Detection](https://mlsanity.com/paper/2508.06203/) (2025)
- [Uni-3DAD: GAN-Inversion Aided Universal 3D Anomaly Detection on Model-free Products](https://mlsanity.com/paper/2408.16201/) (2024)
- [MuSc-V2: Zero-Shot Multimodal Industrial Anomaly Classification and Segmentation with Mutual Scoring of Unlabeled Samples](https://mlsanity.com/paper/2511.10047/) (2025)
- [Back to Point: Exploring Point-Language Models for Zero-Shot 3D Anomaly Detection](https://mlsanity.com/paper/2603.21511/) (2026)
- [AnomalyHybrid: A Domain-agnostic Generative Framework for General Anomaly Detection](https://mlsanity.com/paper/2504.04340/) (2025)

## Abstract (arXiv)

Zero-shot multimodal anomaly detection commonly assumes that RGB and point cloud modalities are equally reliable and can contribute uniformly to anomaly localization. We challenge this assumption. Using a set of recently proposed stringent metrics that penalize false anomaly responses in normal regions, we find that point clouds are substantially more reliable than RGB under zero-shot category shift. Motivated by this observation, we propose WOOPS (When Point Clouds Outperform Pixels), a reliability-aware zero-shot multimodal anomaly detection framework. To strengthen the more reliable geometric modality, we design a Multi-view Information Decoupling module to suppress heterogeneous information from multi-view point cloud projections and enhance point cloud feature quality. To avoid unconditional fusion, we further introduce a Modality Reliability Calibration module to adaptively calibrate modality contributions according to their reliability. Extensive experiments show that our method achieves the best or competitive performance under the new metrics in both unimodal and multimodal settings. Further analysis demonstrates that point cloud information also improves RGB-only inference, while ablations verify the effectiveness of both modules. Code will be released upon acceptance.