Herring populations have been in decline in Eastern Massachusetts. Fish ladders have been installed and populations seem to be starting to recuperate. Scientists have great difficulty counting the juvenile herring as they swim back to sea because of their size and movement patterns. The Great Herring Pond Watershed Association has used crowdfunding to help in their counting efforts, but it’s difficult to find volunteers, and the method is error-prone. We propose a project that uses AI computer vision to create a system that includes the hardware and software that can detect and count the juvenile herrings with high accuracy and speed. This project will be done in conjunction with Florida Atlantic University’s MPCR Lab, University of Massachusetts, Indigo Technologies, Mikolji Corp, and the Great Herring Pond Watershed Association.
It consists of gathering existing video from UMASS and training a computer vision algorithm to detect them. It also contains a hardware component. This project will take around one year. One benefit of this project is that the resulting product can then be licensed by Shoal and sold to other organizations to create a self-sustaining product for Shoal. We already have identified a company that wants to use the technology (Fish Bio).