Fusarium fungi are very common in nature and are often present in soil microbiota. Since some Fusarium species may carry health risks to humans, as well as animal and vegetal species, there is a great interest for detecting them in a reliable way. That is why this paper describes an implementation based both on digital image processing techniques, and artificial neural networks, for identification of some Fusarium species, starting from a microscopic sample image. The subjacent idea is to detect such fungi in an automated fashion, for the sake of eliminating subjective effects and possible human mistakes to the process.