An article co-authored by Susumu Takatsuka and Hidehito Sato, Sony Group Corporation and JAMSTEC* was published in Ecology and Evolution. https://lnkd.in/gPXf8ncD
Ecology and Evolution is an online academic journal published by John Wiley & Sons, a global publisher in science, medicine, and education.
Title: Millisecond-scale behaviours of plankton quantified in vitro and in situ using the Event-based Vision Sensor
This article discusses the outcomes of a novel marine biometric technology research, utilizing an Event-based Vision Sensor (EVS) to capture swimming plankton and estimate the subject by examining features associated with their swimming behaviour from the EVS event data. This research successfully accomplished direct observation in the ocean, a task that was previously challenging. Moving forward, our goal is to develop an observation system capable of monitoring marine ecosystems in real-time and continuously tracking the condition of the world’s oceans.
Compared to conventional image sensors, EVS is suitable for sensing marine particles due to features such as high-speed imaging that accurately captures particle movement and low power consumption achieved through tracking changes in low brightness in the dark ocean and outputting only changes. The study analyzed 22 different features from three perspectives: shape, movement, and periodicity. Previously, white light illumination used in photography was unable to accurately measure natural states because plankton and other organisms exhibit phototaxis, which varies depending on light conditions. EVS, capable of detecting even the smallest changes in near-infrared light at a wavelength of 770 nm imperceptible to most living organisms, enabled non-invasive measurements by utilizing this unique lighting wavelength.
Using these system configurations, we tested a mixture of four types of particles, three types of zooplankton, and fine particles in a laboratory environment. We classified the particle types using machine learning and achieved an accuracy rate as high as 86%. Furthermore, we conducted a world-first experiments on the deep bottoms of Lake Biwa and Suruga Bay through direct underwater observation by EVS. We classified biological and non-biological particles into four groups based on features such as changes in the speed of the particles photographed and the periodicity of changes in particle size due to swimming motion.
We Upon publication of this article, we are offering our proposed observation system that uses EVS to researchers at no cost. We look forward to feedback from the research community.
Part of this research is being conducted in the project selected in the public call of the Ministry of Education, Culture, Sports, Science and Technology in 2021, “Advancement of Technologies for Utilizing Big Data of Marine Life.” Susumu Takatsuka serves as the principal investigator of the project.
* Japan Agency for Marine-Earth Science and Technology