Machine learning to find the primary producers

The ocean is inhabited by a diverse array of single-celled plants called phytoplankton.  Together, they produce half of the oxygen we breathe and provide food for most of the ocean’s creatures.  However, different kinds of phytoplankton play different roles in the marine ecosystem, so it is important to understand how each is distributed. One way to do this is to collect water samples from a place of interest and use imaging techniques to photograph individual phytoplankton cells or groups of cells.  Software is then “trained” to recognize different phytoplankton species in subsequent samples don’t need to be identified manually.  Unfortunately, the machine learning algorithms currently used for this purpose are rather simple and misidentify images most of the time.  Therefore, the goal of this hackathon project would be to develop improved machine learning algorithms to achieve much better accuracy in image identification. This will provide a better picture of the kinds of phytoplankton present in a particular place in the ocean, how many there are, and how might they be functioning in the ecosystem.