PCB DSLR Dataset

The PCB DSLR dataset is meant 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. All images come with accurate PCB segmentation information as well as bounding box information for all Integrated Circuit (IC) chips (9313 samples).

Statistics

Please refer to [1] for detailed statistics.

  • Camera: Nikon D4 with f/2.8 lens and polarization filter
  • Image resolution: 4928×3280 (about 220 ppi)
  • 165 different PCBs, 3 to 5 images per PCB, 748 images in total
  • 2048 unique labeled ICs, 9313 labeled ICs in total
  • 1740 ICs with label information in total

Download

The dataset is split into 8 parts: 1, 2, 3, 4, 5, 6, 7, 8. APIs for convenient use from Matlab, Python2, and C++ are available on github. The dataset is freely available for non-commercial research use. Please cite  [1] when using the dataset.

References

[1] C. Pramerdorfer and M. Kampel: “A Dataset for Computer-Vision-Based PCB Analysis”, Machine Vision Applications (to appear), 2015