SDNET 2025- Defective Bolts and Nuts Steel dataset

The dataset focuses on defective steel connections, specifically missing, loose, and corroded bolts and nuts, along with fatigue cracks. It contains over 826 images of defective and intact bolts captured from both real-world steel structures and laboratory setups. The dataset has been annotated with bounding boxes and masks for deep learning applications in automated defect detection.

本数据集聚焦于钢结构连接件缺陷场景，具体涵盖螺栓螺母缺失、松动、锈蚀以及疲劳裂纹四类典型缺陷。数据集共包含超过826张分别采集自真实钢结构现场与实验室环境的缺陷螺栓与完好螺栓图像。该数据集已完成边界框（bounding boxes）与掩码（masks）标注，可支撑自动化缺陷检测方向的深度学习应用开发。

搜集汇总

背景与挑战

背景概述

SDNET 2025数据集专注于钢铁连接中的缺陷检测，包含826张标注有边界框和掩码的图像，涵盖缺失、松动、腐蚀的螺栓和螺母以及疲劳裂纹。该数据集适用于训练如Faster R-CNN、YOLO和Mask R-CNN等目标检测模型，用于基础设施维护和自动化检测工具的开发。