Identifying prey capture events of a free-ranging marine predator using bio-logger data and deep learning

Royal Society Open Science, Journal Year: 2024, Volume and Issue: 11(6)

Published: June 1, 2024

Marine predators are integral to the functioning of marine ecosystems, and their consumption requirements should be integrated into ecosystem-based management policies. However, estimating prey in diving requires innovative methods as predator–prey interactions rarely observable. We developed a novel method, validated by animal-borne video, that uses tri-axial acceleration depth data quantify capture rates chinstrap penguins ( Pygoscelis antarctica ). These important consumers Antarctic krill Euphausia superba ), commercially harvested crustacean central Southern Ocean food web. collected large set n = 41 individuals) comprising overlapping accelerometer from foraging penguins. Prey captures were manually identified videos, those observations used supervised training two deep learning neural networks (convolutional network (CNN) V-Net). Although CNN V-Net architectures input pipelines differed, both trained models able predict new (linear regression slope predictions against video-observed 1.13; R 2 ≈ 0.86). Our results illustrate algorithms offer means process quantities generated contemporary bio-logging sensors robustly estimate events predators.

Language: Английский

---

Sound evidence for biodiversity monitoring

Science, Journal Year: 2024, Volume and Issue: 385(6705), P. 138 - 140

Published: July 11, 2024

Bioacoustics and artificial intelligence facilitate ecological studies of animal populations

Language: Английский

---

Re-identification of fish individuals of undulate skate via deep learning within a few-shot context

Ecological Informatics, Journal Year: 2023, Volume and Issue: 75, P. 102036 - 102036

Published: Feb. 23, 2023

Individual re-identification is critical to track population changes in order assess status, being particularly relevant species with conservation concerns and difficult access like marine organisms. For this, we propose photo-identification via deep learning as a non-invasive technique discriminate between individuals of the undulate skate (Raja undulata). Nevertheless, accruing enough training samples might be achieve case underwater fish images. We develop novel methodology based on siamese neural network that incorporates statistical fundamentals motivation overcome few-shot context. Our work provides hands-on experience highlights pitfalls when trying apply limited scenario, concerning both data quantity quality, yet providing remarkable results over test set including recaptures, where model capable correctly identifying 70% individuals. The findings this study can strong impact for research teams becoming familiar approaches, it easily extended re-identify other interest from or exploitation point view.

Language: Английский

---

A deep learning model for measuring coral reef halos globally from multispectral satellite imagery

Remote Sensing of Environment, Journal Year: 2023, Volume and Issue: 292, P. 113584 - 113584

Published: April 18, 2023

Reef halos are rings of bare sand that surround coral reef patches. Halo formation is likely to be the indirectly result interactions between relatively healthy predator and herbivore populations. To reduce risk predation, herbivores preferentially graze close safety reef, potentially affecting presence size halo. readily visible in remotely sensed imagery, monitoring their changes may therefore offer clues as how populations faring. However, manually identifying measuring slow limits spatial temporal scope studies. There currently no existing tools automatically identify single measure speed up identification improve our ability quantify variability over space time. Here we present a set convolutional neural networks aimed at from very high-resolution satellite imagery (i.e., ∼0.6 m resolution). We show deep learning algorithms can successfully detect with high degree accuracy (F1 = 0.824), thereby enabling faster, more accurate spatio-temporal halo size. This tool will aid global study halos, ecosystem monitoring, by facilitating discovery ecological dynamics underlying variability.

Language: Английский

---

Applications of machine learning to identify and characterize the sounds produced by fish

ICES Journal of Marine Science, Journal Year: 2023, Volume and Issue: 80(7), P. 1854 - 1867

Published: Aug. 11, 2023

Abstract Aquatic ecosystems are constantly changing due to anthropic stressors, which can lead biodiversity loss. Ocean sound is considered an essential ocean variable, with the potential improve our understanding of its impact on marine life. Fish produce a variety sounds and their choruses often dominate underwater soundscapes. These have been used assess communication, behaviour, spawning location, biodiversity. Artificial intelligence provide robust solution detect classify fish sounds. However, main challenge in applying artificial recognize lack validated data for individual species. This review provides overview recent publications use machine learning, including deep detection, classification, identification. Key challenges limitations discussed, some points guide future studies also provided.

Language: Английский

---

Deep-Learning-Based Automated Tracking and Counting of Living Plankton in Natural Aquatic Environments

Environmental Science & Technology, Journal Year: 2023, Volume and Issue: 57(46), P. 18048 - 18057

Published: May 19, 2023

Plankton are widely distributed in the aquatic environment and serve as an indicator of water quality. Monitoring spatiotemporal variation plankton is efficient approach to forewarning environmental risks. However, conventional microscopy counting time-consuming laborious, hindering application statistics for monitoring. In this work, automated video-oriented tracking workflow (AVPTW) based on deep learning proposed continuous monitoring living abundance environments. With automatic video acquisition, background calibration, detection, tracking, correction, statistics, various types moving zooplankton phytoplankton were counted at a time scale. The accuracy AVPTW was validated with via microscopy. Since only sensitive mobile plankton, temperature- wastewater-discharge-induced population variations monitored online, demonstrating sensitivity changes. robustness also confirmed natural samples from contaminated river uncontaminated lake. Notably, workflows essential generating large amounts data, which prerequisite available data set construction subsequent mining. Furthermore, data-driven approaches pave novel way long-term online elucidating correlation underlying indicators. This work provides replicable paradigm combine imaging devices deep-learning algorithms

Language: Английский

---

Accelerating ocean species discovery and laying the foundations for the future of marine biodiversity research and monitoring

Frontiers in Marine Science, Journal Year: 2023, Volume and Issue: 10

Published: Sept. 27, 2023

Ocean Census is a new Large-Scale Strategic Science Mission aimed at accelerating the discovery and description of marine species. This mission addresses knowledge gap diversity distribution life whereby an estimated 1 million to 2 species between 75% 90% remain undescribed date. Without improved biodiversity, tackling decline eventual extinction many will not be possible. The biota has evolved over 4 billion years includes branches tree that do exist on land or in freshwater. Understanding what ocean where it lives fundamental science, which required understand how works, direct indirect benefits provides society human impacts can reduced managed ensure ecosystems healthy. We describe strategy accelerate rate by: 1) employing consistent standards for digitisation data broaden access biodiversity enabling cybertaxonomy; 2) establishing working practices adopting advanced technologies taxonomy; 3) building capacity stakeholders undertake taxonomic research development, especially targeted low- middle-income countries (LMICs) so they better assess manage their waters contribute global knowledge; 4) increasing observational coverage dedicated expeditions. Census, conceived as open network scientists anchored by Biodiversity Centres developed LMICs. Through collaborative approach, including co-production science with LMICs, funding partners, focus grow current efforts discover globally, permanently transform our ability document, safeguard

Language: Английский

---

The coupling relationship between marine industry agglomeration and ecological environment system from the perspective of ecological environment

Journal of Sea Research, Journal Year: 2024, Volume and Issue: 198, P. 102485 - 102485

Published: March 7, 2024

To analyze the coupling relationship in marine industry agglomeration and ecological environment system, this study used formula of location entropy coefficient for measuring development status industry. Meanwhile, it also separately measures degree primary, secondary, tertiary industries ocean to reveal their region. In addition, a coupled correlation analysis model was designed. It uses grey relational relationships between data series situations insufficient data. Finally, coordination evaluation proposed. evaluates cluster by calculating capacity coefficient, degree, co scheduling. The results show that from 2013 2022, trend Jiaodong Peninsula Province has undergone different changes. primary increased 1.3047 1.0987, while secondary gradually 0.2486 1.1141. R system 0.3986 0.6253. From 2018 increased, moving towards stability coordination. This indicates over past decade, placed greater emphasis on protection, formed positive interaction, promoting coordinated development.

Language: Английский

---

Automated species classification and counting by deep-sea mobile crawler platforms using YOLO

Ecological Informatics, Journal Year: 2024, Volume and Issue: 82, P. 102788 - 102788

Published: Aug. 20, 2024

Edge computing on mobile marine platform is paramount for automated ecological monitoring. The goal of demonstrating the computational feasibility an Artificial Intelligence (AI)-powered camera fully real-time species-classification deep-sea crawler platforms was searched by running You-Only-Look-Once (YOLO) model edge device (NVIDIA Jetson Nano), to evaluate achievable animal detection performances, execution time and power consumption, using all available cores. We processed a total 337 rotating video scans (∼180°), taken during approximately 4 months in 2022 at methane hydrates site Barkley Canyon (Vancouver Island; BC; Canada), focusing three abundant species (i.e., Sablefish Anoplopoma fimbria , Hagfish Eptatretus stoutii Rockfish Sebastes spp.). trained 1926 manually annotated frames showed high test performances terms accuracy (0.98), precision recall (0.99). then applied videos. In 288 videos we detected 133 Sablefish, 31 Hagfish, 321 nearly (about 0.31 s/image) with very low consumption (0.34 J/image). Our results have broad implications intelligent Indeed, YOLO can meet operational-autonomy criteria fast image processing limited energy loads. • Edge-computing allows robots detect, classify count animals situ. An routine tuned operate Wally deep-sea. were Nano, seeking load. Processing sustain autonomy

Language: Английский

---

Accurate Wound and Lice Detection in Atlantic Salmon Fish Using a Convolutional Neural Network

Fishes, Journal Year: 2022, Volume and Issue: 7(6), P. 345 - 345

Published: Nov. 24, 2022

The population living in the coastal region relies heavily on fish as a food source due to their vast availability and low cost. This need has given rise farming. Fish farmers fishing industry face serious challenges such lice aquaculture ecosystem, wounds injuries, early maturity, etc. causing millions of deaths ecosystem. Several measures, cleaner anti-parasite drugs, are utilized reduce sea lice, but getting rid them entirely is challenging. study proposed an image-based machine-learning technique detect presence live salmon farm A new equally distributed dataset contains affected by healthy collected from tanks installed at Institute Marine Research, Bergen, Norway. convolutional neural network for wound detection consisting 15 5 dense layers. methodology test accuracy 96.7% compared with established VGG-19 VGG-16 models, accuracies 91.2% 92.8%, respectively. model false true positive rate 0.011 0.956, 0.0307 0.965 having wounds,

Language: Английский

---