iMagine Empowers Phytoplankton Researchers with Advanced FlowCam Image Analysis

FlowCam phytoplankton identification is the mature use case on which the Flanders Marine Institute (VLIZ) works in iMagine.

The use case focuses on establishing an operational service on the iMagine platform for analysing and processing FlowCam images to determine the taxonomic composition of phytoplankton samples. Phytoplankton plays a crucial role in the aquatic food web, so accurately identifying and classifying these organisms is essential. 

Phytoplankton details seen at the microscope

The technology used for this purpose involves a deep learning image recognition algorithm based on a Convolutional Neural Network (CNN) and a NoSQL MongoDB database. The existing workflow will be enhanced using the iMagine AI platform.

The use case identified various challenges that need to be addressed. These include optimising the data ingestion pipeline, improving metadata and data output formats to comply with community-based standards, enhancing the service to incorporate context input and increasing classification accuracy.

The use case successfully developed its first user story with a model, allowing researchers to reuse a pre-trained classifier or quickly fine-tune the training on their personal training set to optimise the model for their needs.

The model developed, which is already available on the iMagine AI platform, is trained on 350K images from 95 classes of phytoplankton classes and provides the 5 most likely predictions with an overall 98.2% accuracy.

The module will allow researchers to explore the class distribution and model performance by providing notebooks for analysis. 

The annotated dataset used to train the model is available as open access on Zenodo.

Images from the training set