Aqu-AI

AquAI: Monitoring Aquatic Animal Health and Well-Being in Research Facilities

Assessment of aquatic animal health and welfare status in research facilities is commonly based on visual inspection by qualified personnel. However, this type of contact can interfere both with normal and abnormal behavior patterns and reduces the likelihood of early detection of potential problems. For example, fish can get used to human beings and associate their presence with feeding or cleaning activities. Such activities can cause behavioral patterns (e.g. entering a feeding frenzy before the food is applied, or excitation due to expected cleaning process) that also can prevent or mask subtle changes in behavior making it more difficult to estimate health and well-being of the animals.
The AquAI project aims to support aquatic animal research facilities with low-cost hardware and open-access software alternatives to monitor fish behavior continuously as part of the emergency alert system. Facility aquariums are equipped with web-cameras controlled by RaspberryPi-minicomputers to collect images through server-based video surveillance system (VSS) and present them to a machine-learning tool to recognize fish and their swimming behavioral patterns in real-time.

The AquAI project started as a student research project (VetResearch p34-38) and since then parts of the project had been presented at the 25th International Conference on Animal Welfare and the 17th International Conference on Ethology and Animal Husbandry, at VetResearch- retreats and at The 1st International Conference on Science Technology & Innovation – Maejo University (Chiang-Mai, Thailand).

The project is conducted by Dominik von La Roche and Mr. P. Bawidamann (TU Munich), with support from the colleagues in the LMU Veterinary Faculty IT group, and Prof. Dr Dejan Milić (University of Niš, Serbia).