Imaging Seminar: Vision Foundation Models for Microscopy
Event details
| Date | 30.03.2026 |
| Hour | 17:00 › 18:00 |
| Speaker | Constantin Pape, Georg-August-Universität-Göttingen |
| Location | Online |
| Category | Conferences - Seminars |
| Event Language | English |
Registration to attend in person
Zoom link (no registration required)
Abstract: Microscopy image analysis is becoming more important with the increasing size of bioimaging datasets due to continuing improvements spatial and time resolution, field of view, multiplexing, etc. Deep learning-based methods have advanced the state-of-the-art for different analysis tasks, such as cell segmentation, classification, and tracking. However, the large number of different tools for different tasks makes it difficult to find the right solution. Moreover, many current tools suffer from limited generalization capabilities and are thus only applicable in a narrow context. Hence, changes in the data to analyze, e.g. different imaging settings or specimen, require model re-training, which is often time-consuming and technically challenging. Vision foundation models offer a solution to theses challenges by providing a single, powerful model that addresses several image analysis tasks for diverse data conditions.
In this talk, he will present his group’s work on developing vision foundation models for microscopy, with a particular focus on Segment Anything for Microscopy. This approach addresses a wide range of segmentation tasks in both light and electron microscopy within a single framework. He will also introduce ongoing extensions of the model that aim to support additional tasks, including pixel classification, cell classification, and cell tracking, all within a unified system.
Bio: This talk will be given by Prof. Constantin Pape, head of the Computational Cell Analytics research group at the Institute of Computer Science at Georg August University Göttingen and a member of the Campus Institute for Data Science (CIDAS). His research focuses on image processing methods with applications in biology and medicine. Prior to joining Göttingen, he was a postdoctoral fellow at EMBL Heidelberg. His main research interests include machine learning and deep learning, which he applies primarily to microscopy image analysis. He also works on the processing and visualization of large-scale image datasets. His work is strongly interdisciplinary, involving close collaboration with biologists and medical professionals.
The seminar is followed by an aperitif.
Registration appreciated
Zoom link (no registration required)
Abstract: Microscopy image analysis is becoming more important with the increasing size of bioimaging datasets due to continuing improvements spatial and time resolution, field of view, multiplexing, etc. Deep learning-based methods have advanced the state-of-the-art for different analysis tasks, such as cell segmentation, classification, and tracking. However, the large number of different tools for different tasks makes it difficult to find the right solution. Moreover, many current tools suffer from limited generalization capabilities and are thus only applicable in a narrow context. Hence, changes in the data to analyze, e.g. different imaging settings or specimen, require model re-training, which is often time-consuming and technically challenging. Vision foundation models offer a solution to theses challenges by providing a single, powerful model that addresses several image analysis tasks for diverse data conditions.
In this talk, he will present his group’s work on developing vision foundation models for microscopy, with a particular focus on Segment Anything for Microscopy. This approach addresses a wide range of segmentation tasks in both light and electron microscopy within a single framework. He will also introduce ongoing extensions of the model that aim to support additional tasks, including pixel classification, cell classification, and cell tracking, all within a unified system.
Bio: This talk will be given by Prof. Constantin Pape, head of the Computational Cell Analytics research group at the Institute of Computer Science at Georg August University Göttingen and a member of the Campus Institute for Data Science (CIDAS). His research focuses on image processing methods with applications in biology and medicine. Prior to joining Göttingen, he was a postdoctoral fellow at EMBL Heidelberg. His main research interests include machine learning and deep learning, which he applies primarily to microscopy image analysis. He also works on the processing and visualization of large-scale image datasets. His work is strongly interdisciplinary, involving close collaboration with biologists and medical professionals.
The seminar is followed by an aperitif.
Registration appreciated
Links
Practical information
- Informed public
- Registration required