Reinforcement Learning in cv4sensorhub – running
cv4sensorhub is a framework for interdisciplinary image processing algorithms. This system applies reinforcement learning-based algorithms as well. Modern reinforcement learning algorithms uses deep neural networks in the background which requires computationally demanding training.
The goal of our first algorithm is to support semi-automatic image segmentation. The "intelligent drawing tools" can adopt to the characteristics of the given image and this can significantly reduce the number of the required user interventions.
- Project owner:
- Budai Ádám (Automatizálási és Alkalmazott Informatikai Tanszék)
- Web address:
- http://bmeaut.github.io/cv4sensorhub
- Members:
- Automatizálási és Alkalmazott Informatikai Tanszék (VIK-AAIT)
- Cooperations:
In loose cooperation with:
ELTE - Department of Geophysics and Space Science
BME - Department of Construction Materials and Technologies