Acute toxicant warning system based on a fish movement analysis by use of AI concept

Abstract

A fish automonitoring system was developed which enables monitoring for the presence of acute toxicants in water, continuously and quantitatively, based on an image recognition and knowledge-based system. A fish image was first recognized online and then its movement pattern was analyzed. Statistical characteristics of the fish-movement pattern in the water tank were applied to evaluate acute-toxicant presence in the water, based on production rules which were gathered in a series of experiments. The knowledge based was simplified through the aid of image analysis of movement patterns, although the visual observations themselves included complicated information.<>