Coral Intel: a YOLOv8 deep learning framework for monitoring Caribbean corals

Abstract

This study evaluated the performance of YOLOv8 object detection framework for identifying Caribbean scleractinian corals in underwater imagery. A dataset of 10373 coral images, expanded to 32423 images via augmentation, was used to train genus- and species-level models spanning 59 taxa. In the held-out test imagery, the genus-level model achieved a mean average Precision (mAP₅₀) of 0.954, whereas the species-level model reached 0.938, with the performance varying by taxon and generally the highest for morphologically distinctive genera such as Isophyllia and Scolymia. However, independent field validation using transect-based photoquadrat imagery revealed a substantially lower observation-level performance (species-level Precision = 0.389; Recall = 0.336), reflecting the influence of domain shift, morphological similarity among coral taxa, and sparse point-based annotations. Restricting evaluation to the nine species present in the transects improved separability, but did not fully eliminate false positives or missed detections, indicating that label-set reduction does not recover the test set performance under transect survey conditions. The top-down perspective of the field validation imagery lacked key morphological details from oblique-angled training imagery, suggesting model performance is best for near real-time classification from oblique, rather than using top-down, imagery. To facilitate applied deployment for general use, the trained models were integrated into Coral Intel, a web-based framework for automated coral detection currently available in prototype form. Although this study focused on the Caribbean coral taxa, this framework established a transferable foundation for regionally tuned deep learning models that may support scalable reef monitoring.

Keywords

Deep learning / YOLOv8 / Coral reef monitoring / Object detection / Underwater imagery

References

[1] Allemand D, Osborn D. Ocean acidification impacts on coral reefs: from sciences to solutions. Reg Stud Mar Sci, 2019, 28: 100558.

[2] Appeldoorn R, Ballantine D, Bejarano I, Carlo M, Nemeth M, Otero E, et al.. Mesophotic coral ecosystems under anthropogenic stress: a case study at Ponce, Puerto Rico. Coral Reefs, 2016, 35: 63-75.

[3] Beijbom O, Edmunds PJ, Roelfsema C, Smith J, Kline DI, Neal BP, et al.. Towards automated annotation of benthic survey images: variability of human experts and operational modes of automation. PLoS One, 2015, 10(7): e0130312.

[4] Blondin C, Guérin J, Inagaki K, Longo G, Berti-Équille L (2024) Hierarchical classification for automated image annotation of coral reef benthic structures. Preprint at arXiv:2412.08228

[5] Chen QM, Beijbom O, Chan S, Bouwmeester J, Kriegman D (2021) A new deep learning engine for CoralNet. In: 2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). IEEE, pp 3686–3695. https://doi.org/10.1109/ICCVW54120.2021.00412

[6] Cheng S, Wang ZQ, Liu SJ, Han Y, Sun PT, Li JR. Attention-based lightweight YOLOv8 underwater target recognition algorithm. Sensors, 2024, 24: 7640.

[7] Cordero E, Yadav P, Kumar Raju V (2025) SeaMount Geospatial Labs: Coral Intel–AI-powered ocean exploration. https://seamountgeolabs.site/coral_intel. Accessed 5 Aug 2025

[8] da Silveira CBL, Strenzel GMR, Maida M, Gaspar ALB, Ferreira BP. Coral reef mapping with remote sensing and machine learning: a nurture and nature analysis in marine protected areas. Remote Sens, 2021, 13: 2907.

[9] Foreman AD, Duprey NN, Yuval M, Dumestre M, Leichliter JN, Rohr MC, et al.. Severe cold-water bleaching of a deep-water reef underscores future challenges for mesophotic coral ecosystems. Sci Total Environ, 2024, 951: 175210.

[10] Forsman ZH, Barshis DJ, Hunter CL, Toonen RJ. Shape-shifting corals: molecular markers show morphology is evolutionarily plastic in $Porites$. BMC Evol Biol, 2009, 9: 45.

[11] González-Rivero M, Beijbom O, Rodriguez-Ramirez A, Bryant DE, Ganase A, Gonzalez-Marrero Y, et al.. Monitoring of coral reefs using artificial intelligence: a feasible and cost-effective approach. Remote Sens, 2020, 12: 489.

[12] Gress E, Arroyo-Gerez MJ, Wright G, Andradi-Brown DA. Assessing mesophotic coral ecosystems inside and outside a Caribbean marine protected area. R Soc Open Sci, 2018, 5: 180835.

[13] Han L, Zhai JP, Yu ZB, Zheng B. See you somewhere in the ocean: few-shot domain adaptive underwater object detection. Front Mar Sci, 2023, 10: 1151112.

[14] Hedley JD, Roelfsema CM, Chollett I, Harborne AR, Heron SF, Weeks S, et al.. Remote sensing of coral reefs for monitoring and management: a review. Remote Sens, 2016, 8: 118.

[15] Holstein DM, Smith TB, Appeldoorn RS (2016) Ecosystem services provided by mesophotic coral ecosystems. In: Baker EK et al (eds) Mesophotic coral ecosystems–a lifeboat for coral reefs? The United Nations Environment Programme and GRID, Arendal, Nairobi and Arenda, pp 63–66

[16] Humann P, Deloach N. Reef coral identification: Florida, Caribbean, Bahamas, enlarged, 2008, 2, Jacksonville, New World Publications

[17] Jocher G, Chaurasia A, Qiu J (2023) Ultralytics YOLOv8 (version 8.0.0). https://github.com/ultralytics/ultralytics. Accessed Nov 2024

[18] Kleypas JA, Mcmanus JW, Meñez LAB. Environmental limits to coral reef development: where do we draw the line?. Am Zool, 1999, 39: 146-159.

[19] Luz ACN, Barroso VR, Batista D, Lessa AA, Coutinho R, Xavier FC. Early detection of marine bioinvasion by sun corals using YOLOv8. Intell Mar Technol Syst, 2025, 3: 2.

[20] Macusi ED, Katikiro RE, Deepananda KA, Jimenez LA, Conte AR, Fadli N. Human-induced degradation of coastal resources in Asia Pacific and implications on management and food security. J Nat Stud, 2011, 9: 13-28

[21] Pinheiro HT, MacDonald C, Quimbayo JP, Shepherd B, Phelps TA, Loss AC, et al.. Assembly rules of coral reef fish communities along the depth gradient. Curr Biol, 2023, 33: 1421-1430.

[22] Pyle RL. Assessing undiscovered fish biodiversity on deep coral reefs using advanced self-contained diving technology. Mar Technol Soc J, 2000, 34: 82-91.

[23] Rocha LA, Pinheiro HT, Shepherd B, Papastamatiou YP, Luiz OJ, Pyle RL, et al.. Mesophotic coral ecosystems are threatened and ecologically distinct from shallow water reefs. Science, 2018, 361: 281-284.

[24] Rubbens P, Brodie S, Cordier T, Barcellos DD, Devos P, Fernandes-Salvador JA, et al.. Machine learning in marine ecology: an overview of techniques and applications. ICES J Mar Sci, 2023, 80: 1829-1853.

[25] Sannassy Pilly S, Townsend JE, Alisa CAG, Razak TB, Roche RC, Turner JR, et al.. Quantifying coral reef carbonate budgets: a comparison between ReefBudget and CoralNet. Coral Reefs, 2025, 44: 513-527.

[26] Shah C, Nabi MM, Alaba SY, Ebu IA, Prior J, Campbell MD, et al.. YOLOv8-TF: transformer-enhanced YOLOv8 for underwater fish species recognition with class imbalance handling. Sensors, 2025, 25: 1846.

[27] Stoddart J, Jones R, Page C, Marnane M, De Lestang P, Elsdon T. No effect of dredging on the prevalence of coral disease detected during a large dredging program. Mar Pollut Bull, 2019, 140: 353-363.

[28] Trotter C, Griffiths HJ, Whittle RJ. Surveying the deep: a review of computer vision in the benthos. Ecol Inform, 2025, 86: 102989.

[29] Turner JA, Thomson DP, Cresswell AK, Trapon M, Babcock RC. Depth-related patterns in coral recruitment across a shallow to mesophotic gradient. Coral Reefs, 2018, 37: 711-722.

[30] Veron JEN, Stafford-Smith MG, DeVantier LM, Turak E (2016) Corals of the World. https://www.coralsoftheworld.org/page/home/. Accessed 6 Aug 2025

[31] Walker JL, Zeng Z, Wu CCL, Jaffe JS, Frasier KE, Sandin SS. Underwater object detection under domain shift. IEEE J Ocean Eng, 2024, 49: 1209-1220.

[32] Williams B, Balvanera SM, Sethi SS, Lamont TAC, Jompa J, Prasetya M, et al.. Unlocking the soundscape of coral reefs with artificial intelligence: pretrained networks and unsupervised learning win out. PLoS Comput Biol, 2025, 21: e1013029.

[33] Zhang FB, Cao WY, Gao J, Liu SB, Li CY, Song K, et al.. Underwater object detection algorithm based on an improved YOLOv8. J Mar Sci Eng, 2024, 12: 1991.

[34] Zhao YY, Sun FJ, Wu XW. FEB-YOLOv8: a multi-scale lightweight detection model for underwater object detection. PLoS One, 2024, 19: e0311173.

RIGHTS & PERMISSIONS

The Author(s)