MIXED PCB DEFECT DATASET

This dataset contains images focusing on printed circuit boards (PCBs) and their defects. To maintain uniformity, the images are resized to a standard 640 x 640 pixels. To increase the dataset's realism and relevance for real-time sensor applications, intentional augmentation was applied, adding extra defects to simulate real-world scenarios in PCB manufacturing. Various augmentation techniques were used to diversify the dataset, ensuring better algorithm performance during training and evaluation. The dataset features annotations that precisely label the induced defects, such as missing holes, mouse bites, open circuits, shorts, spurs, and spurious copper issues. With its precise annotations and intentional augmentation, this dataset is a valuable resource for advancing research in real-time PCB defect detection and classification. If you used the dataset, please cite the following paper: Kumar Ancha, V., Sibai, F. N., Gonuguntla, V., & Vaddi, R. (2024). Utilizing YOLO Models for Real-World Scenarios: Assessing Novel Mixed Defect Detection Dataset in PCBs. IEEE Access, 12, 100983-100990. https://doi.org/10.1109/ACCESS.2024.3430329本数据集包含聚焦于印刷电路板（Printed Circuit Boards，PCBs）及其缺陷的图像。为维持图像统一性，所有图像均被调整至标准640×640像素尺寸。为提升数据集的真实感与实时传感器应用场景下的适配性，研究人员应用了针对性的数据增强技术，通过人为添加额外缺陷以模拟PCB制造中的真实工业场景，同时通过多样化的增强手段丰富数据集内容，确保算法在训练与评估阶段获得更优异的性能表现。本数据集附带精准标注的人工引入缺陷，涵盖漏孔、咬边、开路、短路、毛刺以及杂散铜缺陷等。凭借精准的标注体系与针对性的数据增强手段，该数据集是推动实时PCB缺陷检测与分类研究的宝贵资源。 若使用本数据集，请引用以下论文： Kumar Ancha, V., Sibai, F. N., Gonuguntla, V., & Vaddi, R. (2024). 面向真实场景的YOLO模型应用：评估PCB新型混合缺陷检测数据集. IEEE Access, 12, 100983-100990. https://doi.org/10.1109/ACCESS.2024.3430329