Marine Ecology Progress Series
MEPS is a leading hybrid research journal on all aspects of marine, coastal and estuarine ecology. Priority is given to outstanding research that advances our ecological understanding.
Impact Factor2.3 (JCR 2026 release)
Article Acceptance Rate51.3% (2025)
Median Time in Review137.5 days (2025)
Total Annual Downloads1.586.389 (2026)
DOI: https://doi.org/10.3354/meps15009
copiedTrawling for photographic bycatch: using computer vision to detect habitat-forming kelp in citizen science fish photographs
Photographs of organisms taken by citizen scientists often contain a wealth of secondary data such as habitat composition information captured in the background. Computer vision has the potential to unlock these secondary data and increase the value of this resource for biodiversity monitoring. Here, we trained a computer vision model using approximately 4000 underwater fish photographs to detect the presence of the common kelp species Ecklonia radiata (hereafter Ecklonia) in the background. The best performing model returned an average area under the precision-recall curve (AUPRC) value of 0.798 on the validation set. The model had an average precision of 0.79 across unseen test sets, meaning that approximately 4 out of every 5 Ecklonia detections were correct. However, the model had a relatively low recall of 0.52 across test sets, meaning it was only detecting half of the fish images that contained Ecklonia. When the best performing model was deployed on an unseen data set of 55000 iNaturalist photographs of fish, approximately 3900 secondary Ecklonia observations were obtained (after accounting for the false positive rate). This represents a 15-fold increase in the number of underwater photographs of Ecklonia in our study region when compared to iNaturalist observations where Ecklonia was the primary subject of the photograph. This study demonstrates how extracting secondary data using computer vision can considerably increase the information obtained from citizen science photographs.
KEYWORDS
ALA (Atlas of Living Australia) (2017) Occurrence record: Herbrecs:BRI AQ1019163. biocache.ala.org.au/occurrences/60575e4f-24c2-4d3c-b4f4-afa23a2a3a12 (accessed 20 September 2024) Link
Beery S (2021) Scaling biodiversity monitoring for the data age. XRDS Crossroads ACM Mag Stud 27:14-18 Crossref
Bewley MS, Nourani-Vatani N, Rao D, Douillard B, Pizarro O, Williams SB (2015) Hierarchical classification in AUV imagery. In: Mejias L, Corke P, Roberts J (eds) Field and service robotics: results of the 9th international conference. Springer, Cham, p 3-16
Blair JD, Gaynor KM, Palmer MS, Marshall KE (2024) A gentle introduction to computer vision-based specimen classification in ecological datasets. J Anim Ecol 93:147-158 Crossref
Bolt MH, Callaghan CT, Poore AGB, Vergés A, Roberts CJ (2022) Using the background of fish photographs to quantify habitat composition in marine ecosystems. Mar Ecol Prog Ser 688:167-172 Crossref
Branson S, Wah C, Schroff F, Babenko B, Welinder P, Perona P, Belongie S (2010) Visual recognition with humans in the loop. In: Daniilidis K, Maragos P, Paragios N (eds) Computer vision—ECCV 2010. ECCV 2010. Lecture Notes in Computer Science, Vol 6314. Springer, Berlin, p 438-451 Crossref
Callaghan CT, Poore AGB, Mesaglio T, Moles AT and others (2021) Three frontiers for the future of biodiversity research using citizen science data. BioScience 71:55-63
Chandra RV, Varanasi BS (2015) Python requests essentials. Packt Publishing Ltd., Birmingham
Costa H, Foody GM, Jiménez S, Silva L (2015) Impacts of species misidentification on species distribution modeling with presence-only data. ISPRS Int J Geoinf 4:2496-2518 Crossref
Davis TR, Champion C, Coleman MA (2021) Climate refugia for kelp within an ocean warming hotspot revealed by stacked species distribution modelling. Mar Environ Res 166:105267 Crossref
DCCEEW (Department of Climate Change, Energy, the Environment and Water) (2023) Integrated marine and coastal regionalisation of Australia (IMCRA) v4.0—Provincial Bioregions. fed.dcceew.gov.au/datasets/erin:integrated-marine-and-coastal-regionalisation-of-australia-imcra-v4-0-provincial-bioregions/about (accessed 5 May 2023) Link
Dilek E, Dener M (2023) Computer vision applications in intelligent transportation systems: a survey. Sensors 23:2938 Crossref
Gildenblat J (2022) Advanced AI explainability with pytorch-gradcam. jacobgil.github.io/pytorch-gradcam-book/introduction.html Link
Kirillov A, Mintun E, Ravi N, Whitehead S and others (2023) Segment Anything. In: Proc 2023 IEEE/CVF International Conference on Computer Vision (ICCV), Paris. p 3992-4003
Koh PW, Sagawa S, Marklund H, Xie SM and others (2021) WILDS: a benchmark of in-the-wild distribution shifts. Proc Mach Learn Res 139:5637-5664
Layton C, Coleman MA, Marzinelli EM, Steinberg PD and others (2020) Kelp forest restoration in Australia. Front Mar Sci 7:74 Crossref
Mahmood A, Giraldo Ospina A, Bennamoun M, An S and others (2020) Automatic hierarchical classification of kelps using deep residual features. Sensors 20:447 Crossref
Marzinelli EM, Williams SB, Babcock RC, Barrett NS and others (2015) Large-scale geographic variation in distribution and abundance of Australian deep-water kelp forests. PLOS ONE 10:e0118390 Crossref
Paszke A, Gross S, Massa F, Lerer A and others (2019) PyTorch: an imperative style, high-performance deep learning library. In: Wallach HM, Larochelle H, Beygelzimer A, d'Alché-Buc F, Fox EB (eds) Proc 33rd Int Conf Neural Information Processing Systems (NeurIPS 2019). Curran Associates, New York, NY, p 8024-8035
Pernat N, Canavan S, Golivets M, Hillaert J and others (2024) Overcoming biodiversity blindness: secondary data in primary citizen science observations. Ecol Solut Evid 5:e12295 Crossref
Pichler M, Hartig F (2023) Machine learning and deep learning—a review for ecologists. Methods Ecol Evol 14:994-1016 Crossref
Pocock MJO, Tweddle JC, Savage J, Robinson LD, Roy HE (2017) The diversity and evolution of ecological and environmental citizen science. PLOS ONE 12:e0172579 Crossref
R Core Team (2023) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna
Roberts CJ, Vergés A, Callaghan CT, Poore AGB (2022) Many cameras make light work: opportunistic photographs of rare species in iNaturalist complement structured surveys of reef fish to better understand species richness. Biodivers Conserv 31:1407-1425 Crossref
Roberts CJ, Vergés A, Poore AGB (2023) A new resource for monitoring reef ecosystems: the background of recreational diver photographs contains valuable habitat data. J Appl Ecol 60:2688-2698 Crossref
Saini M, Susan S (2023) Tackling class imbalance in computer vision: a contemporary review. Artificial Intelligence Rev 56:S1279-S1335 Crossref
Saito T, Rehmsmeier M (2015) The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLOS ONE 10:e0118432 Crossref
Sauder J, Domazetoski V, Banc-Prandi G, Perna G, Meibom A, Tuia D (in press) The Coralscapes Dataset: Semantic Scene Understanding in Coral Reefs. In: Proc 2025 IEEE/CVF Int Conf on Computer Vision (ICCV). Workshops, Honolulu, HI, p 2115-2122
Schneider CA, Rasband WS, Eliceiri KW (2012) NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9:671-675 Crossref
Seeland M, Rzanny M, Boho D, Wäldchen J, Mäder P (2019) Image-based classification of plant genus and family for trained and untrained plant species. BMC Bioinformatics 20:4 Crossref
Shorten C, Khoshgoftaar TM (2019) A survey on Image Data Augmentation for Deep Learning. J Big Data 6:60 Crossref
Strasser BJ, Tancoigne E, Baudry J, Piguet S and others (2023) Quantifying online citizen science: dynamics and demographics of public participation in science. PLOS ONE 18:e0293289 Crossref
Thiel M, Penna-Díaz MA, Luna-Jorquera G, Salas S, Sellanes J, Stotz W (2014) Citizen scientists and marine research: volunteer participants, their contributions, and projection for the future. Oceanogr Mar Biol Annu Rev 52:257-314 Crossref
Tuia D, Kellenberger B, Beery S, Costelloe BR and others (2022) Perspectives in machine learning for wildlife conservation. Nat Commun 13:792 Crossref
Van den Bossche J, Jordahl K, Fleischmann M, McBride J and others (2023) Geopandas/geopandas: v0.14.0. geopandas.org/en/v0.14.0/index.html Link
Vendrow E, Pantazis O, Shepard A, Brostow G and others (2024) INQUIRE: a natural world text-to-image retrieval benchmark. In: Globerson A, Mackey L, Belgrave D, Fan A, Paquet U, Tomczak J, Zhang C (eds) NeurIPS Proceedings: Advances in Neural Information Processing Systems 37 (NeurIPS 2024). Datasets and Benchmarks track. Link
Vergés A, Doropoulos C, Malcolm HA, Skye M and others (2016) Long-term empirical evidence of ocean warming leading to tropicalization of fish communities, increased herbivory, and loss of kelp. Proc Natl Acad Sci USA 113:13791-13796 Crossref
Weinstein BG (2018) A computer vision for animal ecology. J Anim Ecol 87:533-545 Crossref
Wernberg T, Coleman M, Fairhead A, Miller S, Thomsen M (2003) Morphology of Ecklonia radiata (Phaeophyta: Laminarales [sic]) along its geographic distribution in south-western Australia and Australasia. Mar Biol 143:47-55 Crossref
Wernberg T, Bennett S, Babcock RC, de Bettignies T and others (2016) Climate-driven regime shift of a temperate marine ecosystem. Science 353:169-172 Crossref
Wernberg T, Coleman MA, Babcock RC, Bell SY and others (2019) Biology and ecology of the globally significant kelp Ecklonia radiata. Oceanogr Mar Biol Annu Rev 57:265-323 Crossref
Williams J, Coleman MA, Jordan A (2020) Depth, nutrients and urchins explain variability in Ecklonia radiata (Laminariales) distribution and cover across ten degrees of latitude. Aquat Bot 166:103274 Crossref
Womersley HBS (1987) The marine benthic flora of southern Australia, Part II. South Australian Government Printing Division, Adelaide
Wood SN (2011) Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J R Stat Soc Ser B Stat Methodol 73:3-36 Crossref
Christopher J. Roberts (Corresponding Author)
- Centre for Marine Science and Innovation, School of Biological, Earth and Environmental Sciences, UNSW Sydney, Sydney, NSW 2052, Australia
- Centre for Marine Science and Innovation; School of Biological, Earth and Environmental Sciences; UNSW Sydney; Sydney, NSW 2052, Australia
- School of Zoology, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv 69978, Israel
- The Interuniversity Institute for Marine Sciences in Eilat, Eilat 88103, Israel
- California Institute of Technology, Pasadena, CA 91125, USA
Alistair G. B. Poore (Co-author)
- Centre for Marine Science and Innovation; School of Biological, Earth and Environmental Sciences; UNSW Sydney; Sydney, NSW 2052, Australia
Handling Editor:
Lisandro Benedetti-Cecchi, Pisa, Italy
Reviewers:
D. Ayyagari, A. C. Balbar and 1 anonymous referee
Acknowledgements:
We thank the recreational divers and snorkelers who made this research possible by contributing underwater photographs both to the iNaturalist Australasian Fishes project and directly to the research team. We also thank the Resnick Sustainability Institute, the California Institute of Technology, Sara Beery, Pietro Perona and the other lecturers and facilitators of the CV4E Workshop (https://cv4ecology.caltech.edu/), where this computer vision model was initially developed. This research was supported by an Australian Government Research Training Program (RTP) Scholarship to C.J.R. and Australian Research Council Discovery grant 190102030 to A.V.