A dataset for computer-vision-based PCB analysis

We present a public dataset with the aim to facilitate research on computer-vision-based Printed Circuit Board (PCB) analysis, with a focus on recycling-related applications. The dataset contains 748 images of PCBs from a recycling facility, captured under representative conditions using a professional DSLR camera. For all these images we provide accurate segmentation information for the depicted PCBs as well as bounding box information for all Integrated Circuit (IC) chips (9313 samples). Furthermore, we provide textual information of the labels for a subset of 1740 IC samples. By including these different aspects of information, our dataset is useful for designing and testing a variety of methods for PCB analysis, from PCB classification over IC chip localization to the detection of specific chips. We discuss the benefits of PCB analysis for recycling, present dataset statistics, and use the dataset to evaluate two example methods, one for detecting specific PCBs, and one for recognizing mainboards.