The dataset is constructed from images of defective production items that were provided and annotated by Kolektor Group d.o.o.. The images were captured in a controlled industrial environment in a real-world case.

The dataset consists of:

• 399 images:
  • 52 images with visible defects
  • 347 images without any defect
• Original images of sizes:
  • width: 500 px
  • height: from 1240 to 1270 px
• For training and evaluation images should be resized to 512 x 1408 px

For each item, the defect is only visible in at least one image, while two items have defects on two images, which means there were 52 images where the defects are visible. The remaining 347 images serve as negative examples with non-defective surfaces.

DOWNLOAD HERE with fine annotations (for JIM2019 paper)

DOWNLOAD HERE with box annotations (for ICPR2021 and COMIND 2021 papers)

Benchmark results

We split the dataset into three folds to perform 3-fold cross-validation. The splits are available at KolektorSDD-training-splits.zip. The fully prepared TensorFlow dataset split into 3 folds is available at KolektorSDD-dilate=5-tensorflow.zip.

Results on this dataset with our developed network and several state-of-the-art models are available at our research page.

License and citation

The dataset is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. For comerical use please contact Danijel Skočaj.

Please cite our paper published in the Journal of Intelligent Manufacturing when using this dataset:

@article{Tabernik2019JIM,
  author = {Tabernik, Domen and {\v{S}}ela, Samo and Skvar{\v{c}}, Jure and 
  Sko{\v{c}}aj, Danijel},
  journal = {Journal of Intelligent Manufacturing},
  title = {{Segmentation-Based Deep-Learning Approach for Surface-Defect Detection}},
  year = {2019},
  month = {May},
  day = {15},
  issn={1572-8145},
  doi={10.1007/s10845-019-01476-x}
}