AI for HABs: Utilizing Futuristic Tech for Better HABs Mitigation Now

by Maria Beverly Sambajon

Published: MAY 15, 2023

Other

Dr. Paul Ignacio of Team UPB running the AI model specifically developed for the HABs Watch Project

How does the HABs Watch Project apply AI to HABs monitoring?

Deploying the Imaging Flow Cytobot or IFCB in the field is a big help in monitoring phytoplankton activity.It ensures continuous onsite monitoring that can be tedious and time consuming for researchers. However, the IFCB in itself cannot identify possible HABs species. The primary task of the IFCB is to capture images of phytoplankton it detects.

But wait, don’t fret yet. These images,captured by the IFCB can be processed through a convolutional neural network, a type of artificial intelligence model that was developed by the HABs Watch Project UP Baguio team, led by Dr. Paul Ignacio and “trained” to identify HABs species. In order to achieve this, the researchers needed to lay the groundwork by consistently annotating and labeling phytoplankton images and building a working dataset that it will feed to the model. The more expansive the dataset, the more accurate the identification.

Team UPB has completed the preliminary training of different AI models for automated detection and classification of HABs species in the Philippines using open source databases and are now training the model using localized data sets by using images and data collected by other teams within the HABs Watch Project. This allows the model to familiarize itself with actual images of potential HABs species collected by the project and help the model identify images captured by the IFCB in the field more accurately. 

The convolutional neural network model developed for the project is continuously being updated and improved, as part of the HABs Watch Project's objective of real-time monitoring, detection and warning for harmful algal blooms for the health and safety of the Filipino people.