AGSSP数据集
收藏资源简介:
AGSSP数据集是一个包含120,000张工业表面图像的大型数据集,涵盖了61个类别,用于预训练模型。该数据集由北京科技大学钢铁技术协同创新中心收集,其中一半数据来自20个公开可用的工业表面缺陷数据集,另一半来自14个钢铁厂和生产线。该数据集旨在解决金属表面缺陷检测中的数据稀缺问题,并为下游任务提供有效的预训练模型。
The AGSSP dataset is a large-scale dataset comprising 120,000 industrial surface images across 61 categories, intended for model pre-training. Collected by the Collaborative Innovation Center for Iron and Steel Technology, University of Science and Technology Beijing, half of the dataset’s data originates from 20 publicly available industrial surface defect datasets, while the other half is sourced from 14 steel plants and production lines. This dataset aims to address the data scarcity issue in metal surface defect detection, and enables the development of effective pre-trained models for downstream tasks.
- 1Advancing Metallic Surface Defect Detection via Anomaly-Guided Pretraining on a Large Industrial Dataset北京科技大学钢铁技术协同创新中心 · 2025年



