Deep Learning for Image Analysis

Europe/Berlin
European Neuroscience Institute Göttingen

European Neuroscience Institute Göttingen

ENI-G, Grisebachstr. 5, 37077 Göttingen
Constantin Pape (UMIN), Antonio Politi (Facility for Light Microscopy, MPI-NAT, Göttingen), Martin Schilling (UKEI), Luca Freckmann (UMIN)
    • 9:00 AM 12:30 PM
      Classification
      • 9:00 AM
        Lecture: Intro to DL for microscopy, classification and PyTorch 1h 30m
      • 10:30 AM
        Getting started 2h
        • Participant introductions
        • Setting up JupyterHub server
        • Introducing first exercise on classification
    • 12:30 PM 1:30 PM
      Lunch 1h
    • 1:30 PM 5:00 PM
      Classification
      • 1:30 PM
        First exercise: Cell classification 3h
      • 4:30 PM
        Recap first exercise 30m
    • 9:00 AM 12:30 PM
      Segmentation
      • 9:00 AM
        Lecture: Segmentation, pre-trained models, transfer learning 1h 30m
      • 10:30 AM
        Intro to exercises for Nucleus and Cell segmentation 2h
    • 12:30 PM 1:30 PM
      Lunch 1h
    • 1:30 PM 5:00 PM
      Segmentation
      • 1:30 PM
        Second exercise: Cell segmentation 2h 30m
      • 4:00 PM
        Recap exercise 2. Image restoration and lecture recap. 1h
    • 9:00 AM 12:30 PM
      Other deep learning tools
      • 9:00 AM
        Lecture: Overview of deep learning tools (micro-sam, cellpose, stardist, nn UNet, trackstra, ...) 1h 30m
      • 10:30 AM
        Intro to third exercise 2h
    • 12:30 PM 1:30 PM
      Lunch 1h
    • 1:30 PM 5:45 PM
      Other deep learning tools
      • 1:30 PM
        Third exercise: Group work where we try a tool per group 2h 30m
      • 4:00 PM
        Recap third exercise, project discussion 1h
    • 9:00 AM 12:30 PM
      Project work
    • 12:30 PM 1:30 PM
      Lunch 1h
    • 1:30 PM 5:00 PM
      Project work
    • 9:00 AM 12:30 PM
      Project work
    • 12:30 PM 1:30 PM
      Lunch 1h
    • 1:30 PM 3:00 PM
      Project work
    • 3:00 PM 5:00 PM
      Final discussion

      Wrap-up and how to continue to use the infrastructure