Deep neural networks outperform human expert's capacity in characterizing bioleaching bacterial biofilm composition.

•Deep learning has become widely used in different fields of computer science such as face recognition, but also in biology, for example to detect malignant skin cancers based on images.•Deep learning applied to microscopy images of biofilm colonization patterns accurately characterized its bacterial composition.•Deep learning applications only rarely outperform human experts in classification tasks. Here however, deep learning reached an accuracy of 90%, therefore clearly outperforming human experts (50% accurate).•Our method provides an accurate alternative to standard, time-consuming biochemical methods, using visual information only.