Deep Learning has changed the game in Computer Vision on a fundamental level and images from „ordinary“ camera applications (traffic, social media etc) can now be interpreted algorithimcally with impressive sensitivity and precision. However, in the domain of bioimage informatics (microscopy, MALDI imaging, hyperspectral imaging, histopathology and marine imaging) the application of deep learning architectures seems less straight forward as some deep learning prerequisites are not met (lower data volume, lesser training labels, strong class imbalance, lower signal quality, weak labels etc). In this seminar we want to explore the recent works published on this matter and discuss the open issues.
Rhythmus | Tag | Uhrzeit | Format / Ort | Zeitraum |
---|