We are a group of marine, social & data scientists as well as system engineers who all care about the protection of marine life. The organisations developing the SUBSIM infrastructure include University of Gothenburg, Wildlife.ai, Combine, and Swedish Geological Survey.

by Paulina Izabela de Jesús Álvarez, University of Gothenburg

Eelgrass meadows have been declining widely in the past due to diseases and climate change, causing a substantial loss of ecosystem services. Current efforts to restore and maintain this key habitat depend on effective monitoring methods. In this study we develop semi-automated monitoring methods based on underwater drones and computer vision. Our results show that deep learning models are able to reliably identify presence-absence signals of seagrass meadows and in addition recognize the presence of alternative vegetation types. The results are summarised in the recent MSc thesis with the title "Studying biodiversity and distribution of benthic vegetation with deep learning methods." (de Jesús Álvarez 2025), which you can find under Publications.

by Louis Fiorina, University of Gothenburg and Andreas Sundelöf, Swedish University of Agricultural Sciences

Fishing for European Lobster (Homarus gammarus) is an economically significant activity in many coastal regions of Europe. Understanding lobster populations, along with their behavior and responses to fishing pressure is key to improving fisheries’ sustainability. Lobster stock assessments are likewise important for measuring the performance of Marine Protected Areas (MPAs). By using baited traps fitted with cameras, we can integrate video technology with machine learning and develop an effective way to monitor both lobster stocks and lobster behavior. In this project, we combine Baited Remote Underwater Video systems (BRUVs) with computer vision-based object detection models to classify and monitor European Lobster individuals, assess size class distribution in local populations, and movement patterns, as well as capture details of lobster capture dynamics. This project, based on footage collected around the Kåvra MPA on the west coast of Sweden, is a collaboration between scientists at the University of Gothenburg and the Swedish University of Agricultural Sciences. Results are summarised in the recent MSc thesis (Fiorina 2025), which you can find under Publications.

by Matthias Obst, University of Gothenburg

SUBSIM has launched a new collaboration with SINTEF-Norway. Here we join forces and develop real-time AI/ML fish identification and tracking methods on video streams. The goal of the collaboration is to establish automated monitoring services for continuous assessment of status and change in macroplanktonic communities, including fish, jellyfish and large crustaceans. The methods will likewise be useful to assess the impact of new species becoming established in a region. Currently, we do have video observatories in the Trondheim fjord and Oslofjord (Norway) as well as in the Kosterhavets National park (Sweden). The project will deliver biological data and analytical tools to Digital Twin of the Ocean, through the projects ILIAD and DTO-bioflow.

Find out more about this project

Iliad pilot 03 - Environmental monitoring - Fish identification and tracking

Ensemble Learning for Robust Fish Species Identification Across Diverse Underwater Environments

by Christian Nilsson, University of Bergen

Vertical zonation in marine environments occurs when sessile species are primarily found at specific depth intervals, dividing the habitat into distinct horizontal layers. The community assemblage and size of these different zones is determined by environmental conditions as well as interactions between organisms. This phenomenon has primarily been studied in intertidal waters, but also occurs at greater depths, beneath the halocline. Here, low spatiotemporal variation in environmental conditions suggest increased ecosystem sensitivity to recent climate change. However, subtidal environments and their zonation patterns have been understudied historically due to technical limitations. Therefore, in this study we apply machine learning based object detection software to ROV footage to estimate and quantify vertical zonation and community structure along the sides of the Koster submarine canyon. We also plan to investigate possible changes in zonation structure as well as any possible correlations to changing environmental conditions over the past 30 years. For more information - have a look at Christian's latest bloggpost, this conference poster, or his manuscript in review.

by Xhoni Dalipi, University of Gothenburg

The recent emergence of aquacultures has gained significant attention in recent years due to its potential to address food security and environmental concerns. Seaweed and shellfish cultivation offer advantages such as providing a sustainable source of protein, reducing pressure on wild fisheries, and contributing to nutrient uptake in marine environments. However, these practices also raise concerns about their impact on marine biodiversity. Alterations to local ecosystems and potential habitat modification are some of the challenges that need to be addressed. This study will provide a comprehensive understanding of the relationship between small-scale aquaculture and benthic biodiversity, by recording biological communities across the farming season in small aquacultures on the Swedish West coast. The findings of this research can potentially contribute to the design of sustainable and nature-positive aquacultures in the future. The model developed by the project can be downloaded under Publications. 

by Frida Persson, University of Gothenburg

In this project our student went to South Africa to sample shark observations and population density data, together with the Sharklife Conservation Group. We now have the first models trained to detect up to six shark species in the marine footage recorded with baited remote underwater video systems (BRUVS). The models may be used in the future to discover new and continuously monitor existing elasmobranch hotspots that need special protection from fisheries. The full report is available under Publications. 

by Leon Green, University of Gothenburg

Scientists at the University of Gothenburg are investigating the interaction between invasive and native fish on the Swedish west coast. A large part of the data collection takes place through underwater video recording and will be used to streamline the analysis of the many hours of video recordings. "Invasions from alien species are expected to increase in the future, so it is absolutely critical that we learn how to use such new image analysis services" says Leon Green who studies how biodiversity can limit the spread of alien species. He continues: "...especially the combination big data methods and citizen science can become an important toolkit for mapping the distribution and population growth of invasive fish.”

by Saga Alsterlind and  Louis Fiorina, University of Gothenburg

As the global demand for green energy continues to rise, so does the market for offshore wind farms (OWF). The implementation of the submerged OWF foundations introduces new hard substrates on which marine organisms begin to colonize, also known as the “reef effect”. OWF's therefore have great potential to increase marine biodiversity by elevating habitat value. This project aims to develop a method for quantifying and monitoring changes of sessile biodiversity in relation to OWF's, utilizing machine-learning to perform segmentation detection on still imagery. This project is a collaboration between University of Gothenburg and the renewable energy company OX2. Report are available under Publications. 

by Rasmus Torslund, University of Gothenburg

Sea pens are coral animals that live on soft bottoms and act as important habitat builders when they grow dense. Such sea pen aggregations create shelter for fish and shellfish. Previous studies have shown that sea pen habitats are in decline due to harmful fishing methods such as bottom trawling but also from other threats. By gaining an understanding of the distribution and abundance of sea pens and how this abundance relates to trawling activities we hope to provide insight into the status and changes of soft bottom ecosystems in the Kattegat region. For more information - have a look at the report under Publications.

by Joel Hjärne Kokk, University of Gothenburg

In this project we investigated the efficiency and reliability of the machine learning (ML) approach compared to manual counting of habitat building species. We trained and tested the performance of an object detection model including three species: Desmophyllum pertusum (cold water coral), Bolocera tuediae (deplete sea anemone) and Geodia barretti (football sponge). We then used the model to see whether the distribution of these three species had changed over the last twenty years in one of the reef areas of Kosterhavet National Park. The results highlight the potential of ML models as an analytical tool, but also identify issues that need to be addressed to make the system applicable for large-scale studies on deep water fauna such as better standardization of subsea imagery. The mapping tool developed during the project is available here.

by Victor Anton, Wildlife.ai

California’s marine protected areas (MPAs) are ocean and coastal areas with special protections to help conserve and protect marine ecosystems and wildlife. As MPAs are predominantly underwater, monitoring them and their effects can be challenging. By leveraging the power of community scientists and machine learning technology this project aims to identify and extract classified images of organisms from underwater videos for contribution to biodiversity databases. Ultimately, these observations will integrate with ongoing efforts by the California Academy of Sciences and MPA managing agencies to use community science data to better understand and monitor biodiversity across California’s coast and MPAs. You can engage with the project here.

by Sarah Al-Khateeb and Lisa Bodlak

Large and diverse data is crucial to train object detection models properly and achieve satisfactory prediction performance. However, in some cases, for example when developing models for detecting rare species data will always be limited resulting in overfitting and poor model performance. Furthermore, underwater images can often be of poor quality, with colour distortion and blurriness. In this study we explored methods to both increase and enhance training data for object classes with little training data and thereby improve detection accuracy. We experimented with several Generative Adversarial Networks (GANs) to enhance and increase the training data. The results prepared for publication and the synthetic images are published here.

by Victor Anton, Wildlife.ai

Marine reserves are the highest form of protection for Aotearoa New Zealand’s marine environment, facilitating the recovery of exploited species. To ensure their effectiveness, the New Zealand Department of Conservation monitors the state and trend of exploited species using baited underwater videos. Baited underwater videos are a versatile underwater survey method to compare the abundance of carnivorous and scavenger fishes inside and outside marine reserves. The team is using SUBSIM tools to process the videos from the 44 marine reserves. Read the latest news here.

Lorem ipsum dolor sit amet, ius error audiam at. Duo ne mentitum posidonium. At quando aliquam scripserit cum. Ferri postea mandamus no vix. Ei sed modo omittam, te primis iudicabit hendrerit his.

Ancillae nominati contentiones ius cu, et quo altera indoctum vituperata. Putent fuisset duo id. Ne sit alii habeo invidunt, ad qui dolores adolescens, qui id hinc tollit scripta. Pro cu velit eleifend, sea cu erat persius, duo ad agam scripta laboramus. Nam aperiam detracto intellegat ea, quas oratio invenire sed ei.

Munere legere dissentias eam eu. Cu sensibus dignissim contentiones sit. Habemus constituam sea et, illud recusabo sed id. Te vim nostrum perpetua euripidis, in per habemus antiopam gubergren. Scaevola interpretaris eu ius, id sit movet apeirian. Ad soleat legere quo. Adipisci dignissim reformidans cum an.

Ius at commodo malorum, an vis graece aliquam contentiones. Eros dicant integre qui ne, est cu illum labores electram. Eam id omnis temporibus. An mutat error mei, te sit erant singulis inimicus. Qui movet utroque vituperata no, pro no mollis aperiam scribentur.

Lorem ipsum dolor sit amet, ius error audiam at. Duo ne mentitum posidonium. At quando aliquam scripserit cum. Ferri postea mandamus no vix. Ei sed modo omittam, te primis iudicabit hendrerit his.

Ancillae nominati contentiones ius cu, et quo altera indoctum vituperata. Putent fuisset duo id. Ne sit alii habeo invidunt, ad qui dolores adolescens, qui id hinc tollit scripta. Pro cu velit eleifend, sea cu erat persius, duo ad agam scripta laboramus. Nam aperiam detracto intellegat ea, quas oratio invenire sed ei.

Lorem ipsum dolor sit amet, ius error audiam at. Duo ne mentitum posidonium. At quando aliquam scripserit cum. Ferri postea mandamus no vix. Ei sed modo omittam, te primis iudicabit hendrerit his.

Ancillae nominati contentiones ius cu, et quo altera indoctum vituperata. Putent fuisset duo id. Ne sit alii habeo invidunt, ad qui dolores adolescens, qui id hinc tollit scripta. Pro cu velit eleifend, sea cu erat persius, duo ad agam scripta laboramus. Nam aperiam detracto intellegat ea, quas oratio invenire sed ei.

Munere legere dissentias eam eu. Cu sensibus dignissim contentiones sit. Habemus constituam sea et, illud recusabo sed id. Te vim nostrum perpetua euripidis, in per habemus antiopam gubergren. Scaevola interpretaris eu ius, id sit movet apeirian. Ad soleat legere quo. Adipisci dignissim reformidans cum an.

Ius at commodo malorum, an vis graece aliquam contentiones. Eros dicant integre qui ne, est cu illum labores electram. Eam id omnis temporibus. An mutat error mei, te sit erant singulis inimicus. Qui movet utroque vituperata no, pro no mollis aperiam scribentur.