Abstract: An AI-based system and method for feed monitoring in a land-based fish farm, where the fish farm has a tank containing fish and an effluent pipe that is coupled to the tank where uneaten feed pellets and non-pellet objects flow through the effluent pipe to an exit port. A feed camera is mounted to an effluent pipe for capturing a video feed having images of objects that traverse a field of view of the feed camera when each image is acquired. A special-purpose computer executes a pellet-tracking algorithm that employs a region of interest (ROI) proposal module, an ROI classification module, an ROI tracking module, and a trajectory classification module for at least counting uneaten feed pellets. The computer generates in real time a pellet count based on the pellet trajectories.