Sign in Subscribe
Natural resources

In-Situ Online Monitoring Systems and Artificial Intelligence Aid Coral Reef Ecosystems

Strategic Insights

— 2 min read

Current research from the Ministry of Natural Resources has highlighted the importance of coral reefs, which are rapidly degrading globally. In-situ online monitoring systems are being used to monitor coral reef ecosystems in real-time, enabling the study of coral reef ecology through artificial intelligence technology, particularly deep learning technology. A new dataset, SCSFish2025, has been proposed to aid in the automatic detection and identification of coral reef fish, which is crucial for the study of marine biodiversity and ecosystems.

Key Takeaways:

  • Coral reefs are one of the most biodiverse ecosystems on Earth, but are rapidly degrading globally, necessitating the use of in-situ online monitoring systems to study coral reef ecology in real-time.
  • Artificial intelligence technology, particularly deep learning technology, is playing an increasingly important role in the study of coral reef ecology, especially in the automatic detection and identification of coral reef fish.
  • The proposed dataset, SCSFish2025, contains 11,956 high-resolution underwater surveillance images and over 120,000 bounding boxes covering 30 species of fish, manually labelled by experienced fish identification experts.
  • The SCSFish2025 dataset establishes a benchmark for the detection performance of deep learning object detection techniques, with the best baseline model RT-DETRv2 achieving mAP@50 performance of 0.9960 and 0.7486 respectively on the five-fold cross-validation of the training set and the independent test set.
  • The release of this dataset will help promote the development of AI technology in the study of automatic detection and identification of coral reef fish, and provide strong support for the study of marine biodiversity and ecosystems.
  • The project code and dataset are available at https://github.com/FudanZhengSYSU/SCSFish2025.

Statistics:

  • 11,956 high-resolution underwater surveillance images are present in the SCSFish2025 dataset.
  • Over 120,000 bounding boxes covering 30 species of fish are present in the SCSFish2025 dataset.
  • The SCSFish2025 dataset was manually labelled by experienced fish identification experts.
  • The best baseline model RT-DETRv2 achieved an mAP@50 performance of 0.9960 and 0.7486 respectively on the five-fold cross-validation of the training set and the independent test set.
  • The dataset is designed for the study of marine biodiversity and ecosystems in the waters of China's Nansha Islands.

Sources:

  • SCSFish2025: a large dataset from South China sea for coral reef fish identification. Scientific Reports, 2025,15(1):1-17. (Scientific Reports - http://www.nature.com/srep/index.html)
  • Studies from Ministry of Natural Resources Update Current Data on Technology (SCSFish2025: a large dataset from South China sea for coral reef fish identification). Journal of Engineering. September 1, 2025; p 488.

Read more