Forefront | AI technology protects endangered species, voiceprint recognition provides “identity cards” for Hainan gibbons

Text | Angie Li

Editor | Su Jianxun

AI technology is making people more and more understand the voice of Hainan gibbons. Recently, the Hainan National Park Research Institute and Huawei held a tropical rainforest visit in Hainan Province, presenting the habitat environment of the world-class rare animal Hainan gibbon to the public, and at the same time discussing how technology can help biodiversity conservation.

It is understood that the Hainan gibbon is listed as a “critically endangered” species by the IUCN (World Conservation Union), which is even more endangered than the giant panda. There are only 36 remaining, all of which are distributed in more than ten Bawangling Mountains in Hainan Tropical Rainforest National Park. In a square kilometer forest. Due to the inability to cultivate artificially, the population growth of gibbons completely depends on habitat restoration and population protection. At present, its habitat has expanded to more than 4,000 mu since 2003.

It is precisely because of the vast habitat that it is difficult to monitor the Hainan gibbons. The monitoring team patrols the deep mountains of Bawangling all year round, and often encounters problems such as difficult mountain roads, high temperature and rain, and mosquitoes infestation. But more importantly, Hainan gibbons are alert by nature and live in trees all year round. Direct observation is very difficult. Judgment can only be monitored by the chirping of the gibbons.

Source: Thematic Science Exhibition on International Gibbon Day 2022

According to Wei Fuliang, a member of the Hainan gibbon Bawangling monitoring team, the Hainan gibbon gets up and sings every morning, and the monitoring team can also collect and analyze the sound of the Hainan gibbon by deploying monitoring equipment. In the past, monitoring work generally relied on manual methods. Due to the limited storage space of offline devices, data can generally only be stored for 15 days, and monitors can only patrol the same monitoring site every 3 months. Sounds are collected normally.

Starting from the end of 2021, combining digital technologies such as connectivity, cloud, and AI, Huawei has joined forces with IUCN and the Hainan National Park Research Institute to create a pilot Tech4Nature project to explore how digital technologies can bring new solutions to biodiversity conservation.

Source: Huawei

Through the wireless communication network provided by Huawei’s partners, 24-hour real-time monitoring can now be achieved, and data can be sent back to the cloud, breaking away from the traditional mode of regularly extracting offline device data on-site and greatly improving the real-time nature of data acquisition.

With the help of cloud computing and artificial intelligence, the efficiency and accuracy of data processing have also been significantly improved, and many gibbon calls that are difficult to identify manually can be successfully extracted and analyzed. Prof. Fan Pengfei from Sun Yat-Sen University used algorithms to extract sound features, and through a series of model fitting, he has achieved a recognition accuracy of 89.2%.

Huawei said that in the future, through Huawei’s AI development framework, it is expected to realize automatic identification and classification of gibbon calls, establish a unique “sound ID card” for each gibbon, achieve “individualized” accurate analysis, and help animal protection experts judge the gibbon family. Membership of gibbons to develop more targeted programs for the conservation of gibbons.