NFDI4BIOIMAGE
Published June 25, 2025 | Version v1
Presentation Open

Vision Language Models for Bio-image Data Science

Authors/Creators

  • 1. ROR icon Center for Scalable Data Analytics and Artificial Intelligence
  • 2. ROR icon Leipzig University

Description

In this talk, I demonstrate potential use-cases for vision-language models (VLM) in bio-image data science, focusing on how to analyse microscopy image data. It covers these use-cases:

  • cell counting
  • bounding-box segmentation
  • image descriptions
  • VLMs guessing which algorithm to use for processing
  • Data analysis code generation
  • Answering github issues 

The talk also points at a number of VLM-based open-source tools which start reshaping the scientific bio-image data science domain:

Files

Haase_LLMs_BIA_Helmholtz_Imaging.pdf

Files (42.4 MB)

Name Size Download all
md5:9b6fcb1394dd2ea2d4c098f4ff29b0f6
4.3 MB Preview Download
md5:93e4dfaab83e92725bf147b537ed73c6
38.1 MB Download