PCB-IND: Photovoltaic PCB Defect Dataset for YOLO, VOC, and COCO Formats

The PCB-IND dataset contains 4846 industrial PCB images annotated for 8 defect types: mouse_bite, missing_copper, scratch, spurious_copper, copper_burr, stain, short, and open. Images are split into train (3876 images), val (485 images), and test (485 images) sets. The dataset is provided in YOLO, VOC, and COCO formats for deep learning applications in defect detection. Each image is 300x300 pixels. The dataset is licensed under CC BY 4.0.