Title : Novel information technologies for fish behavior recognition
Abstract:
There is an increasing recognition within the aquaculture industry that understanding the behavior of farmed animals can help provide solutions to farming problems. Behavior monitor would help farmers to observe animals behavior as welfare indicators for the better and effective management of aquaculture facilities. These behaviors including feeding, mating, swimming and abnormal behavior of aquatic animal during cultivation may help to reflect the growth status and achieve accurate predictions of the water quality environment. Until now, aquatic animals behavior identification has been mostly based on manual monitoring, which is usually inaccurate, time-consuming and laborious. Until now, machine vision, acoustics technology, mathematical models and sensors provide the possibility of developing automatic, faster and cheaper methods for behavior recognition of farmed animals in aquaculture. Machine vision technology provides an automated, non-invasive, cost-effective method to record behavioral parameters; acoustic waves undergo little propagation loss in water, and their propagation distances are long, making them the best way to detect and identify small objects underwater; accelerometer has enabled long-term, real-time monitoring a range of physiological and behavioral variables that are either directly or indirectly relevant to framed animals health and productivity. This paper also forecast several different trends in farmed animal behaviors monitor to further improve the level of precision farming: 1) Using information fusion technique will be applied in dead zones where single equipment is inaccessible; 2) developing and expanding the function of underwater sensors, multi-functional sensor can detect multiple information of fish at the same time; 3) using a rapid, accurate deep learning algorithm to monitor behaviors based on images and videos. These future directions will have great significance to accelerate the development of new means and techniques for more effective behavior monitor. However, the accuracy and intelligence still need to be improved to meet intensive aquaculture requirements. Through close cooperation between fisheries experts and engineers, the precision and the level of intelligence for farmed animal behaviors monitor will be further improved based on the above methods.
