Collaborative Vision-Based Precision Monitoring of Tiny Eel Larvae in a Water Tank

This paper proposes a novel monitoring system in which two cameras installed outside the water tank collaborate to photograph eel larvae for tracking and analysis. Long-term and periodic observations of fish can provide important information about their life. However, fresh eel larvae have a tiny and transparent body, making it difficult to observe and track them through optical vision. To address this problem, we proposed a monitoring system that uses a fixed high-resolution camera to observe the entire breeding tank and a telecentric zoom camera on a Cartesian robot to track and observe a single larva. In addition, the collaborative vision-based object search method helps the zoom camera to capture photographs of the tiny larvae. We verified the method by placing several 3D plastic models similar to eel larvae in the water tank at predetermined locations. We then conducted an experiment using actual larvae and were able to obtain videos of the larvae. Subsequently, we analyzed the videos to obtain the population estimation, shape, and size of the larvae. As a result, we established a novel eel larva monitoring system and conducted actual larva-monitoring experiments.