Plankton are often overlooked creatures, but not by the researchers at Engineers for Exploration who partnered with the Jaffe Lab at Scripps Institute of Oceanography for the study of these small creatures. These tiny animals are the foundation of the entire marine ecosystem. Without them, the ocean would cease to function; we would have no whales, coral reefs, or even fish.

With that knowledge, it is extremely important that we closely study plankton, and how humans affect them. But this is a task more easily said than completed. Plankton, by nature, are hard to study, especially because of there size. For decades, people have been studying them by sampling with a fine net. But this rudimentary process has severe limitations, especially in the amount of data for the work put in.

Diving off Scripps pier with camera

This is where E4E comes in. We worked closely with SIO in developing algorithms to process the data from their newly developed underwater plankton microscope. Such a system was developed to fully utilize the potential of a permanent microscope in collecting massive amounts of data. We developed machine learning algorithms to interface with a meteor web app. This system allows for the crowd sourcing of information about the plankton, which in turn retrains the algorithm, which then in turn assists the user in classification.

Classification Graphs
Classification Graphs
CAD Model of Plankton Camera

Overall, this system allows for rapid classification of plankton, thus allowing researchers to correlate important water conditions and other factors to the abundance of certain classes of plankton. An important step in further understanding these curious little creatures, which in turn is an important step in protecting our oceans.

Video feed from Plankton Cam

Check out the current web app at https://jaffeweb.ucsd.edu:5001 (site is temporarily down, please see https://jaffeweb.ucsd.edu) and keep an eye out for an iOS app coming soon.

Diagram of Plankton Camera workflow
Diagram of Plankton Camera workflow