Knowledge Commons of Institute of Automation,CAS
ESTATE: Expert-Guided State Text Enhancement for Zero-Shot Industrial Anomaly Detection | |
Bingke Zhu1![]() ![]() | |
2024 | |
会议名称 | IEEE International Conference on Image Processing |
会议日期 | 2024.10.27-2024.10.30 |
会议地点 | Abu Dhabi, UAE |
摘要 | The Expert-Guided State Text Enhancement Anomaly Detection (ESTATE) framework addresses the challenges in industrial anomaly detection arising from diverse product categories and limited defective samples. This framework, integrating expert insights through comparative state prompts, leverages two innovative text-guided networks, CLS-Refner and SEG-Refner, enhancing model training. These networks, connected to residual textual features of standard vision-language pre-trained models, focus on amplifying adjectives’ signifcance in text for improved image block and pixel-level alignment. ESTATE’s effectiveness is demonstrated through evaluations on MVTecAD and VisA datasets, achieving AUROC scores of 89.6%/89.6% for classifcation and 95.1%/85.0% for segmentation tasks, alongside setting new benchmarks in F1Max and PRO metrics.The AUC-cls on MVTecAD and VisA demonstrated an enhancement of 5.06% and 8.97%, respectively, compared to the APRIL-GAN approach. |
收录类别 | EI |
七大方向——子方向分类 | 图像视频处理与分析 |
国重实验室规划方向分类 | 视觉信息处理 |
是否有论文关联数据集需要存交 | 否 |
文献类型 | 会议论文 |
条目标识符 | http://ir.ia.ac.cn/handle/173211/57459 |
专题 | 紫东太初大模型研究中心 |
通讯作者 | Hao Li |
作者单位 | 1.Foundation Model Research Center, Institute of Automation, Chinese Academy of Sciences, Beijing, China 2.School of Computer Science and Engineering, Central South University, Hunan, China 3.School of Instrument Science and Opto-electronics Engineering, Hefei University of Technology, Hefei, China |
第一作者单位 | 中国科学院自动化研究所 |
推荐引用方式 GB/T 7714 | Bingke Zhu,Hao Li,Changlin Chen,et al. ESTATE: Expert-Guided State Text Enhancement for Zero-Shot Industrial Anomaly Detection[C],2024. |
条目包含的文件 | ||||||
文件名称/大小 | 文献类型 | 版本类型 | 开放类型 | 使用许可 | ||
ESTATE__Expert_guide(2871KB) | 会议论文 | 开放获取 | CC BY-NC-SA | 浏览 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论