We are pleased to announce our contribution to the Data Curation and Augmentation in Medical Imaging Workshop (DCAMI) at the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2024 in Seattle, USA. Congratulations to the authors!
Distribution-Aware Multi-Label FixMatch on CheXpert. Sontje Ihler, Felix Kuhnke, Timo Kuhlgatz, Thomas Seel
The focus of the work is Semi-Supervised Learning (SSL) for the diagnosis of X-ray images. SSL is the combined learning from annotated and non-annotated data and a good strategy for the automation of medical image analysis, since often large amounts of images are available, but it is too expensive and time-consuming to have them all annotated by medical professionals. Unfortunately, the field of multi-label SSL has so far been insufficiently researched. Our team has succeeded in extending the probably most popular SSL method (FixMatch) for multi-label problems and thus presenting a very simple and therefore promising multi-label SSL strategy.
We would like to thank the entire team, especially Sontje, for their dedicated work on this project. Your collaboration and hard work has played a crucial role in this success.
You can find more information here:
https://dca-in-mi.github.io/
https://cvpr.thecvf.com/