[ON-SITE] Vision models use cases with AI
Training room 207, IT4Innovations
ON-SITE
Annotation
The goal of this course is to teach participants how to prepare their own training data primarily for image segmentation, possibly also for image classification, and to show them creating of an own model of a neural network that specifically addresses their own image data and task. Throughout the course, basic terms and theory will be explained and, above all, shared practical experience will be provided through short, light-weight lectures.
All work will rely solely on SW tools with open licenses that do not exclude commercial use. The principal tool will be Fiji, a graphical desktop application (an ordinary program with a window, operated with a mouse) for image processing. This is a well-known program that is also continuously developed towards remote and large data, and that can be found on essentially every biologist's desktop in scientific and academic settings. Within Fiji, we will use the Labkit plugin, which allows images to be annotated and segmented, either manually by drawing with the mouse or semi-automatically, AI-assisted. Labkit offers only neural networks, such as Cellpose or Segment-Anything-Model, that work in isolation (no user data leaves a computer), and are relatively undemanding computationally. Finally, we will demonstrate a simple Python module for training new models and show how to bring such a model back into Labkit, as well as how to use it directly, whether locally or remotely.
Benefits for the attendees
Attendees will learn how to:
- Prepare and annotate their own image datasets for segmentation and classification.
- Use Fiji and Labkit for manual and AI-assisted image segmentation.
- Train a neural network tailored to their own images and research tasks.
- Deploy trained models in Labkit or use them independently, both locally and remotely.
- Build privacy-conscious workflows using open-source tools suitable for commercial use.
The course combines essential theory with hands-on guidance and practical experience.
Level
Intermediate
Language
English
Prerequisites
The course assumes participants primarily from biology, medicine, and related fields.
Technical requirements
- Data to work on will be freely available, though participants are welcome to bring their own data.
- Participants shall bring their own laptops. Power plugs will be available, operating systems Windows, Linux, and Macs are all supported, 5 GB of disk space shall be available for the course.
- The laptops need to be set up prior to the course according to the provided video instructions.
- Basic computer literacy is sufficient for creating annotations.
Tutor
Dr. Vladimír Ulman is a computer scientist specializing in biomedical image processing and analysis, with a focus on large-scale bioimage data. He earned his PhD in computer science in 2011. Currently, he holds appointments at the IT4Innovations National Supercomputing Center (VSB – Technical University of Ostrava) and at CEITEC, Masaryk University, in Brno. He develops open-source tools mainly within the Fiji/ImageJ ecosystem and gives hands-on training in bioimage analysis, including Fiji/ImageJ, Napari, and AI tools for image processing whenever opportunities arise.
LUMI AI Factory is funded jointly by the EuroHPC Joint Undertaking, through the European Union's Connecting Europe Facility and the Horizon 2020 research and innovation programme, as well as Finland, the Czech Republic, Poland, Estonia, Norway, and Denmark.
This course was supported by the Ministry of Education, Youth and Sports of the Czech Republic through the e-INFRA CZ (ID:90254).
.png)
All presentations and educational materials of this course are provided under the Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license.