The way microbial analysis of samples is performed has not changed much in most
microbiology labs for the past 50 years. This research attempts to utilize computer vision for the automation of microbiological analysis. A common way of automation is to provide tools to a domain expert so they can confi gure the automated system. This inadvertently requires the domain expert to have technical knowledge. To solve this issue several end-user trainable systems based on evolutionary algorithms and neural networks are proposed. These automatically configured systems have been shown to outperform manually configured systems.
Masters Thesis
Klaas Dijkstra
