2026-11-17 Deep Learning on VIBE (09:00-17:00)
Reiter
Deep Learning on VIBE
This one-day, hands-on course introduces deep-learning software on the Virtual Imaging in Bern (VIBE) platform. VIBE is a virtual desktop running on the university's UBELIX HPC cluster. In your browser you get GPU computing and ready-to-use applications such as Cellpose, napari, QuPath and Fiji, with nothing to install. Most examples come from microscopy, but the course is also relevant to scientists working with other kinds of imaging data.
The course covers:
- Introduction to deep learning for microscopy: core concepts, machine learning vs neural network, model assessment via standard metrics.
- Introduction to VIBE: accessing the platform, launching applications and managing data.
- Segmentation with Cellpose: segment live-cell images, correct the predictions and retrain a model in the GUI, assess quality with standard metrics, and reuse your model in QuPath.
- Denoising with CAREamics: train a CARE model in napari on low- and high-signal-to-noise image pairs, follow the loss curves, and export your model.
Course Objectives
- Understand the basics of deep learning and best practices for image analysis
- Train, fine-tune and evaluate models for segmentation and denoising images
- Become familiar with VIBE for your own analysis pipelines
Target group
- All UniBE members, but in particular PhD students and postodcs who need to develop bioimage analysis pipelines. No prior deep-learning experience is needed.
Prerequisites
- Participants must bring their own laptops
Methods
The course will alternate between presentation of topics and practical exercises.
Course material
- Introduction to Deep Learning: https://docs.google.com/presentation/d/11k_3zv1_x3OpMwn8JHIblmiupifjR_5LE89TI1hd-24/edit?usp=sharing
- Introduction to VIBE: https://docs.google.com/presentation/d/14cbuepglgaIbQRVxJMnp0LSI_G3kP_SWPycECnw1588/edit?usp=sharing
- VIBE documentation: https://dsl-unibe-ch.github.io/vibe-documentation/
Certificate
- A certificate will be delivered to participants who have attended the whole training.
Coaches
- Ruben Lopez is a research software engineer at the MIC of the University of Bern, specialised in bioimage analysis and application specialist for VIBE
- Guillaume Witz is a research software engineer at the DSL of the University of Bern, specialised in bioimage analysis.
Time : 2026-11-17, 09:00-17:00
Location : Room 116, Muesmattstrasse 27, 3012 Bern (Muesmatt Library Building)
Training language: English
Participants : Max 14
Registraion : Mandatory
The Data Science Lab is there to boost your research by supporting you solving computing challenges.
https://www.dsl.unibe.ch/