# Investigating the Semantic Robustness of CLIP-based Zero-Shot Anomaly Segmentation

arXiv: https://arxiv.org/abs/2405.07969 · HTML: https://mlsanity.com/paper/2405.07969/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 30 references that are in the map.
- First posted on arXiv 13 May 2024.

## Builds on (most cited first, 22 of 22)

- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computation](https://mlsanity.com/paper/1308.3432/) (2013)
- [Natural Adversarial Examples](https://mlsanity.com/paper/1907.07174/) (2019)
- [Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs](https://mlsanity.com/paper/2401.06209/) (2024)
- [SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation](https://mlsanity.com/paper/2207.14315/) (2022)
- [AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection](https://mlsanity.com/paper/2310.18961/) (2023)
- [Kornia: an Open Source Differentiable Computer Vision Library for PyTorch](https://mlsanity.com/paper/1910.02190/) (2019)
- [AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models](https://mlsanity.com/paper/2308.15366/) (2023)
- [Registration based Few-Shot Anomaly Detection](https://mlsanity.com/paper/2207.07361/) (2022)
- [Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models](https://mlsanity.com/paper/2402.12336/) (2024)
- [Does CLIP Bind Concepts? Probing Compositionality in Large Image Models](https://mlsanity.com/paper/2212.10537/) (2022)
- [Backdoor Attacks for In-Context Learning with Language Models](https://mlsanity.com/paper/2307.14692/) (2023)
- [Semantic Adversarial Examples](https://mlsanity.com/paper/1804.00499/) (2018)
- [Unrestricted Adversarial Examples via Semantic Manipulation](https://mlsanity.com/paper/1904.06347/) (2019)
- [Anomalib: A Deep Learning Library for Anomaly Detection](https://mlsanity.com/paper/2202.08341/) (2022)
- [On Diffusion Modeling for Anomaly Detection](https://mlsanity.com/paper/2305.18593/) (2023)
- [Segment Any Anomaly without Training via Hybrid Prompt Regularization](https://mlsanity.com/paper/2305.10724/) (2023)
- [ImageNet-X: Understanding Model Mistakes with Factor of Variation Annotations](https://mlsanity.com/paper/2211.01866/) (2022)
- [A Closer Look at the Robustness of Contrastive Language-Image Pre-Training (CLIP)](https://mlsanity.com/paper/2402.07410/) (2024)
- [Towards Generic Anomaly Detection and Understanding: Large-scale Visual-linguistic Model (GPT-4V) Takes the Lead](https://mlsanity.com/paper/2311.02782/) (2023)
- [Effective Robustness against Natural Distribution Shifts for Models with Different Training Data](https://mlsanity.com/paper/2302.01381/) (2023)
- [Visual Data-Type Understanding does not emerge from Scaling Vision-Language Models](https://mlsanity.com/paper/2310.08577/) (2023)

## Abstract (arXiv)

Zero-shot anomaly segmentation using pre-trained foundation models is a promising approach that enables effective algorithms without expensive, domain-specific training or fine-tuning. Ensuring that these methods work across various environmental conditions and are robust to distribution shifts is an open problem. We investigate the performance of WinCLIP [14] zero-shot anomaly segmentation algorithm by perturbing test data using three semantic transformations: bounded angular rotations, bounded saturation shifts, and hue shifts. We empirically measure a lower performance bound by aggregating across per-sample worst-case perturbations and find that average performance drops by up to 20% in area under the ROC curve and 40% in area under the per-region overlap curve. We find that performance is consistently lowered on three CLIP backbones, regardless of model architecture or learning objective, demonstrating a need for careful performance evaluation.