Til hovedinnhold

Recent Advances in Visual Sensing and Machine Learning Techniques for in-situ Plankton-taxa Classification

Recent Advances in Visual Sensing and Machine Learning Techniques for in-situ Plankton-taxa Classification

Kategori
Konferansebidrag og faglig presentasjon
Sammendrag
The assessment of planktonic organisms is a prevailing task in marine ecology and oceanography as planktonic organisms form the principal food source for consumers at higher trophic levels. Reliable estimates on the production at the lowermost trophic levels are thus an integral part for the management of marine ecosystems. Traditional plankton sampling and analysis are limited in their spatial and temporal context of the organisms’environment, which are often critical clues to a biologist for its habitation. In addition, ship-based sampling as described leads to a high level of uncertainty in the estimation, since point measurements that are intermittent in space and time are used (Vannier, 2018; Lermusiaux, 2006; Reid et al., 2003). A disruptive change in approach to tackle this problem is currently taking place, enabled by the use of autonomous robots augmented by visual sensing for real-time analysis (Ohman et al., 2019; Sosik and Olson, 2008; Roberts and Jaffe, 2007; Henthornet al., 2006).With the recent advances in deep learning, enabled by the computational power of multi-core CPUs and General-Purpose GPUs, processing and classification of large datasets while learning higher level representations,is now possible. Enhanced traditional machine learning methods are driven by multiple kernel learning, wheregeneral features are combined with robust features and new types from multiple views are defined in order to generate multiple classifiers (Py et al., 2016; Dai et al., 2016; Lee et al., 2016; Moniruzzaman et al., 2017). In this paper, we present recent DL methods for microscopic organisms’ identification and classification. We recommend the use of a small DL architecture based on our performance metrics comparison which reported an accuracy of 95% as opposed to approximately(90% - 93%)achieved by the state-of-the-art networks: ZooplanktoNet, VGGNet, AlexNet, ResNet, and GoogleNet, while training over a labeled dataset of extracted objects from images of plankton organisms captured in-situ. The selected DL architecture is embedded into alight-weight autonomous vehicle (LAUV) for real-time in-situ plankton-taxa identification and classification.The LAUV in turn utilizes the feedback from the real-time image analysis to constantly update a probability density map that further enables an adaptive sampling process.
Oppdragsgiver
  • Norges forskningsråd / 262701
Språk
Engelsk
Forfatter(e)
Institusjon(er)
  • Norges teknisk-naturvitenskapelige universitet
  • SINTEF Ocean / Miljø og nye ressurser
Presentert på
Ocean Sciences Meeting 2020
Sted
San Diego, CA
Dato
15.02.2020 - 20.02.2020
Arrangør
AGU, ASLO and TOS
År
2020