Saved in:
Bibliographic Details
Main Authors: Ueno, Shiryu, Hayashi, Yoshikazu, Nakatsuka, Shunsuke, Yamada, Yusei, Aizawa, Hiroaki, Kato, Kunihito
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.09057
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915149623853056
author Ueno, Shiryu
Hayashi, Yoshikazu
Nakatsuka, Shunsuke
Yamada, Yusei
Aizawa, Hiroaki
Kato, Kunihito
author_facet Ueno, Shiryu
Hayashi, Yoshikazu
Nakatsuka, Shunsuke
Yamada, Yusei
Aizawa, Hiroaki
Kato, Kunihito
contents We propose general visual inspection model using Vision-Language Model~(VLM) with few-shot images of non-defective or defective products, along with explanatory texts that serve as inspection criteria. Although existing VLM exhibit high performance across various tasks, they are not trained on specific tasks such as visual inspection. Thus, we construct a dataset consisting of diverse images of non-defective and defective products collected from the web, along with unified formatted output text, and fine-tune VLM. For new products, our method employs In-Context Learning, which allows the model to perform inspections with an example of non-defective or defective image and the corresponding explanatory texts with visual prompts. This approach eliminates the need to collect a large number of training samples and re-train the model for each product. The experimental results show that our method achieves high performance, with MCC of 0.804 and F1-score of 0.950 on MVTec AD in a one-shot manner. Our code is available at~https://github.com/ia-gu/Vision-Language-In-Context-Learning-Driven-Few-Shot-Visual-Inspection-Model.
format Preprint
id arxiv_https___arxiv_org_abs_2502_09057
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Vision-Language In-Context Learning Driven Few-Shot Visual Inspection Model
Ueno, Shiryu
Hayashi, Yoshikazu
Nakatsuka, Shunsuke
Yamada, Yusei
Aizawa, Hiroaki
Kato, Kunihito
Computer Vision and Pattern Recognition
We propose general visual inspection model using Vision-Language Model~(VLM) with few-shot images of non-defective or defective products, along with explanatory texts that serve as inspection criteria. Although existing VLM exhibit high performance across various tasks, they are not trained on specific tasks such as visual inspection. Thus, we construct a dataset consisting of diverse images of non-defective and defective products collected from the web, along with unified formatted output text, and fine-tune VLM. For new products, our method employs In-Context Learning, which allows the model to perform inspections with an example of non-defective or defective image and the corresponding explanatory texts with visual prompts. This approach eliminates the need to collect a large number of training samples and re-train the model for each product. The experimental results show that our method achieves high performance, with MCC of 0.804 and F1-score of 0.950 on MVTec AD in a one-shot manner. Our code is available at~https://github.com/ia-gu/Vision-Language-In-Context-Learning-Driven-Few-Shot-Visual-Inspection-Model.
title Vision-Language In-Context Learning Driven Few-Shot Visual Inspection Model
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2502.09057