Marine plankton play important role in ocean productivity and carbon cycling, and are sensitive to anthropogenic impacts ranging from coastal seawater eutrophication, pollution, ocean acidification, to global climate change. In particular, analysis of planktonic organism is indispensable for HAB study and monitoring. Therefore, their observation and quantitation is of fundamental significance for oceanographic research and coastal environment monitoring.

However, current marine plankton observation still relies heavily on traditional manual water sampling and optical microscopy, which has long been notoriously slow and labor intensive. Molecular methods such as qPCR has very good species-specificity, but they are usually slow and can hardly give reliable cell abundance information. In situ fluorimeters can work underwater continuously for long time, which can measure relative concentration fluctuation of chlorophyll a as a proxy for HAB warning, but they lack of taxonomy capability. Remote sensing can provide global view of large area sea surface, which has greatly deepened and widened our understanding about the generation and elimination mechanisms of HAB via ocean color analysis. But remote sensing can hardly obtain subsurface information and needs in situ data for model validation.

In the information era, it is widely believed that machine vision can be very promising to aid people for more efficient, objective, and reproducible in situ plankton observation and HAB species monitoring at individual organism level. This has led to the development of in situ optical imaging instruments and relevant machine learning algorithms. In situ imaging instruments eliminate the need for plankton sample collection, fixation and preprocessing. With the continuous development of artificial intelligence and the decrease of computation performance-to-cost ratio, machine vision is enabling HAB and plankton study a unique opportunity for improving observational resolution and scale in the dimensions of time, space, and taxonomy with better efficiency.

In this talk, the presenter will introduce the basics, challenges, and progresses of the technology, and will also discuss the potential of artificial intelligence for future development and applications of the in situ plankton observation methods.