IMIS

Pre-trained Phytoplankton Species Classifier (VLIZ) Deep Learning Model for the Identification of 95 Phytoplankton Species This repository provides a pre-trained deep learning model for species-level classification of phytoplankton, developed by the Flanders Marine Institute (VLIZ). The model is designed to work with high-resolution images captured using FlowCam technology and can identify 95 phytoplankton species using a convolutional neural network (CNN) architecture. Three different model variants are included in this repository: Exception Model (TensorFlow 1) – An earlier implementation using TensorFlow 1. Exception Model (TensorFlow 2) – A re-implementation of the same architecture using TensorFlow 2. Phytoplankton_EfficientNetV2B0.tar – The best-performing model, based on the EfficientNetV2B0 architecture, also implemented in TensorFlow 2. The classes used in training are added as txt file. This model can be used to predict phytoplankton species or to re-train a new CNN within the repo. For my information, take a look at our GitHub GitHub - ai4os-hub/phyto-plankton-classification:  https://github.com/ai4os-hub/phyto-plankton-classification