Lightweight automatized species and media-invariant colony counting application

Abstract

Colony counting is a routine procedure in the microbiology field, which is labor-intensive, time-consuming, and prone to human-introduced variations, which can hinder assay accuracy. A simple way to overcome this issue is to use an automated computer system capable of performing fast, accurate, and repeatable colony counting. Historically, for this task, the two most common techniques used were simple edge detection, and generic convolutional neural networks, capable of image processing. In this article, we present a simple desktop application based on a lightweight, accurate deep learning model fine-tuned for colony counting under different conditions.