Vision-Based Real-Time Monitoring on the Behavior of Fish School

This paper introduces a technique which can automatically monitor the behavior of fish school in images in real-time. Results on the activity level, distribution and social interaction within the school are generated based on the spatial information extracted from captured images. The fish behaviors we observe here are selected based on a list of responses which fish exhibit when they are in distress. As it is a very challenging task to perform manual observations on fish, this technique creates a convenient alternative for researchers who need to study the behavior of fish. Instead, monitoring can be done effortlessly as images are translated to statistical results which can be used to describe the behavior of fish in the school. On top of this, the results can also detect any change to the water quality.