Beyond medical imaging

Image processing and (weakly) supervised learning in other datasets

Description

Next to medical imaging occasionally lab members work on other types of data, with methods often at least partially inspired by (traditional) image processing and weakly-supervised learning techniques. Some projects include (not all are published yet):

  • Nikolaj Kjøller Bjerregaard developed a method for recognizing furigana in Japanese texts and comic books
  • Jannik Elsässer did a project on recognizing sea grass in underwater images
  • Patrick Wittendorff Abarzua Neira, Silas Roien Arildsen, Bertram Kosmo Hviid and Tobias Skovbæk Brund did a project on machine learning for sign language recognition
  • Dovile did a project on predictive maintenance of ball bearings in the pharmaceutical industry

Credits

The cover image is by Becca Walthall from Noun Project (CC BY 3.0)

References

  1. furigana.png
    Detection of Furigana Text in Images
    Nikolaj Kjøller Bjerregaard , Veronika Cheplygina, and Stefan Heinrich
    arXiv preprint arXiv:2207.03960, 2022
    URL
  2. pharmaceutical.png