Monitoring modern-day fisheries is extremely challenging. Every day, millions of vessels harvest the oceans, and many remain unregulated, putting target and non-target fish populations at risk of overexploitation and collapse. To combat this, fisheries managers are increasingly turning to Electronic Monitoring (EM) technologies â integrated systems of cameras and sensors â to record onboard activity and gather the large amounts of data required to make sound regulatory decisions. Fisheries managers require advanced data analysis tools to handle the surplus of data produced by these systems. A single commercial fishing trip can produce over 1000 hours of video, which must be manually reviewed on-land to quantify all fishing activities, including fish that are caught and discarded, and encounters with protected species. Due to the substantial time required for processing, without significant technological advancements these technologies are forecast to cover only 1.5% of the worldâs fishing activity by 2030. This project will build upon recent research in artificial intelligence and computer vision to automate fisheries data processing, making EM technologies viable for widespread adoption. Specifically, Ai.Fish will prototype a specialized computer vision system for counting and classifying fish onboard commercial fishing vessels using video and sensor data from EM sys
