Investigating aquatic ecosystems with computer vision, machine learning and the internet of things

Abstract

This paper describes an independent research project undertaken by a pair of high school students in Singapore under the mentorship of a Research Scientist at the National Institute of Education. The importance of Artificial Intelligence (AI) and Machine Learning (ML) continues to increase every day in our constantly advancing society. This investigative study aims to uncover the practicality of both AI and ML by utilizing it in a real world context, more specifically, to monitor aquatic behaviour. The aim of this investigation focuses on how ML can be used alongside appropriate hardware to effectively monitor the behaviour of aquatic organisms in response to changes in environmental factors. This paper describes the design, construction, testing and design decisions behind the development of a module which allows us to use ML in tandem with IoT sensors to fulfill the above aim. This investigation ultimately concludes that we were able to successfully conceptualize and create two designs and accompanying prototypes where their strengths lie in the presence of an ecosystem in which sensor data measured can be easily compared to the number of fishes detected by our object detection model to draw relationships between aquatic behaviour and different environmental factors. However it is limited due to the main hindrance of the prototypes’ incapability of maintaining clear visibility of fish in murky waters with high turbidity.