Living benthic foraminifera are used as environmental proxies to evaluate the quality of marine ecosystems. This evaluation is usually based on diversity indices and/or on a group of indicative species with specific ecological requirements (tolerant to sensitive species). For this purpose, sorting and identification of living benthic foraminifera are needed which is time-consuming and requires taxonomical expert knowledge. In this study, we present an approach to automatically identify living (Rose Bengal stained) benthic foraminifera using the "ParticleTrieur" software (Marchant et al. 2020) and test its applicability for biomonitoring purposes. Samples from the intertidal mudflat of the Atlantic coast and the French Mediterranean coast were photographed by an automated machine consisting of a 3D printer moving in X, Y, and Z space and a camera connected to an objective with a ring light. Focus stacked images were always taken with the same exposure, stack height, and stack step. Through initial manual segmentation in the "Computer Vision Annotation Tool" (CVAT) and later training in the segmentation by artificial intelligence, living benthic foraminifera are cropped as individual images from full-field images. Using "ParticleTrieur", foraminifera images are manually labelled at the species level and a model is trained to automatically classify future images of interest. Convolutional Neural Network training is performed using Tensorflow libraries. The trained models will be applied to unclassified datasets to compare human and artificial intelligence classification concerning different ecological indices in both contrasted study areas.