TIPS Over Tricks: Simple Prompts for Effective Zero-shot Anomaly Detection

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Salehi, Alireza, Karami, Ehsan, Noey, Sepehr, Noey, Sahand, Yamada, Makoto, Hosseini, Reshad, Sabokrou, Mohammad
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911468231852032
author Salehi, Alireza
Karami, Ehsan
Noey, Sepehr
Noey, Sahand
Yamada, Makoto
Hosseini, Reshad
Sabokrou, Mohammad
author_facet Salehi, Alireza
Karami, Ehsan
Noey, Sepehr
Noey, Sahand
Yamada, Makoto
Hosseini, Reshad
Sabokrou, Mohammad
contents Anomaly detection identifies departures from expected behavior in safety-critical settings. When target-domain normal data are unavailable, zero-shot anomaly detection (ZSAD) leverages vision-language models (VLMs). However, CLIP's coarse image-text alignment limits both localization and detection due to (i) spatial misalignment and (ii) weak sensitivity to fine-grained anomalies; prior works compensate with complex auxiliary modules yet largely overlook the choice of backbone. We revisit the backbone and use TIPS-a VLM trained with spatially aware objectives. While TIPS alleviates CLIP's issues, it exposes a distributional gap between global and local features. We address this with decoupled prompts-fixed for image-level detection and learnable for pixel-level localization-and by injecting local evidence into the global score. Without CLIP-specific tricks, our TIPS-based pipeline improves image-level performance by 1.1-3.9% and pixel-level by 1.5-6.9% across seven industrial datasets, delivering strong generalization with a lean architecture. Code is available at github.com/AlirezaSalehy/Tipsomaly.
format Preprint
id arxiv_https___arxiv_org_abs_2602_03594
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TIPS Over Tricks: Simple Prompts for Effective Zero-shot Anomaly Detection
Salehi, Alireza
Karami, Ehsan
Noey, Sepehr
Noey, Sahand
Yamada, Makoto
Hosseini, Reshad
Sabokrou, Mohammad
Computer Vision and Pattern Recognition
Anomaly detection identifies departures from expected behavior in safety-critical settings. When target-domain normal data are unavailable, zero-shot anomaly detection (ZSAD) leverages vision-language models (VLMs). However, CLIP's coarse image-text alignment limits both localization and detection due to (i) spatial misalignment and (ii) weak sensitivity to fine-grained anomalies; prior works compensate with complex auxiliary modules yet largely overlook the choice of backbone. We revisit the backbone and use TIPS-a VLM trained with spatially aware objectives. While TIPS alleviates CLIP's issues, it exposes a distributional gap between global and local features. We address this with decoupled prompts-fixed for image-level detection and learnable for pixel-level localization-and by injecting local evidence into the global score. Without CLIP-specific tricks, our TIPS-based pipeline improves image-level performance by 1.1-3.9% and pixel-level by 1.5-6.9% across seven industrial datasets, delivering strong generalization with a lean architecture. Code is available at github.com/AlirezaSalehy/Tipsomaly.
title TIPS Over Tricks: Simple Prompts for Effective Zero-shot Anomaly Detection
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.03594