Microalgae classification based on machine learning techniques

Abstract

In this paper, two models for classification of microalgae species based on artificial neural networks have been developed and validated. The models work in combination with FlowCAM, a device capable of capturing each of the particles detected in a sample and obtaining a set of descriptive features for each one. One of the models uses these feature variables as input, while the other makes use of the captured images, both being able to distinguish between two well-known species of microalgae, Scenedesmus almeriensis and Chlorella vulgaris, calculating the proportion of each of these in the analyzed mixture. The models were trained with pure samples of each specie and validated using mixed combinations of them. The results confirm the potential of image analysis and deep learning techniques for the identification of microalgae cultures, as well as the higher accuracy of the feature-based model, thus extending the range of classification approaches in this field.