AI-Powered Coral Conservation in China's Dongshan Island
Coral reefs are believed to have the highest biodiversity of any ecosystem on the planet. Despite occupying less than 1% of the ocean floor, coral reefs are home to around 25% of all marine life, including 4,000 species of fish.
However, temperature rises caused by climate change, the excessive growth of plant and algae, and overfishing are some of the factors threatening the survival of reef ecosystems.
The Dongshan Coral Reserve near Dongshan Island is home to China's northernmost hermatypic coral community. Unlike other coral communities around the country, the Dongshan community is particularly sensitive to changes in hydrodynamic conditions and climate fluctuations in the climate due to its unique geographical location and ecological characteristics. Monitoring and analyzing the Dongshan coral communities provides valuable insights that can contribute to coral conservation by enabling scientific research and the development of robust conservation strategies, alongside education programs for the broader community, including school children and local fishermen.

Hermatypic corals
Researchers at Xiamen University are performing long-term monitoring of the coral reefs in Dongshan. Video data of coral and fish collected by underwater cameras are transmitted to data transfer stations on the nearest island via optical cables, and then to the experiment station over wireless networks. Researchers then analyze this data to identify different species of corals and fish. This long-term, continuous online monitoring project has generated huge numbers of images and videos of marine life in the Dongshan Coral Reserve, laying a solid foundation for refined data analysis.
Coral conservation powered by AI and digital technologies
In the past, researchers had to manually analyze underwater video data, which was both difficult and inefficient. The quality of these videos was heavily affected by factors like weather and water quality.
Fish species identification
Coral species identification (marked in different colors)
Now, with the rapid development of AI, work efficiency has been greatly improved. An intelligent coral and fish identification and analysis system powered by Huawei’s Ascend AI has been developed through the joint efforts of multiple organizations, including Huawei's TECH4ALL team, Xiamen University's research team, the Society of Entrepreneurs and Ecology Foundation, and the Fujian AI Computing Center.

Dongshan Island Biodiversity Monitoring Platform
This system analyzes and processes the underwater video data collected by Xiamen University in real time to quickly identify coral and fish species and record the quantity and size of fish swarms, generating coral and fish information libraries that includes images and videos. Currently, the identification accuracy for the five hermatypic coral species is 99% and the average identification accuracy for 35 fish species is 93%. It previously took researchers two or three days to analyze a 5-hour video, but this is now possible in just 40 minutes thanks to AI.
-
99%
identification accuracy for coral
-
7 x
increase in identification accuracy
The intelligent monitoring project in the Dongshan Coral Reserve demonstrates the huge potential of digital technologies in nature conservation.
Huawei TECH4ALL is committed to applying digital technologies, such as 5G, optical networks, cloud, and AI, and bringing in additional partners to help conservationists and researchers conserve nature with technology in more protected areas.