CHENGDU/LANZHOU, June 10 (Xinhua) -- At a wildlife rescue and breeding base in Wenxian County, northwest China's Gansu Province, a screen tracks wild giant pandas in real time. Base director Liu Zhi can plan the next patrol route with just a few clicks -- a far cry from over two decades ago, when rangers trudged through mountains guided only by luck and experience.
Liu has been working in the mountains since 2000. According to him, the area managed by the base is one of the regions with the highest density of wild giant pandas in the Giant Panda National Park, which was officially established in 2021 to strengthen protection of giant pandas and their habitats, spanning the three provinces of Sichuan, Shaanxi and Gansu.
"Many wild animals, including giant pandas, are actually quite shy. One of our tasks is to protect wildlife and their habitats while minimizing human impact on nature," Liu said.
A "space-air-ground" intelligent monitoring system is now in place locally, integrating satellite remote sensing, aerial photography, and ground sensor networks. With more than 1,500 infrared cameras and over 20 drones deployed, and linked to a smart forestry and grassland big data platform, the system enables round-the-clock, multidimensional monitoring of hydrology, meteorology, wildlife and plant life.
"The new technologies give us a more accurate understanding of population changes," Liu added.
Similar tech-driven efforts are now quietly unfolding deep in the forests of other areas within the Giant Panda National Park. In the area of Chengdu, the capital of Sichuan Province, more than 600 infrared cameras have been deployed, capturing over 300 images of wild giant pandas each year on average. With 178 monitoring grids also set up, an initial intelligent ecological monitoring system has thereby been established, combining biodiversity monitoring, electronic fencing, and smart patrolling.
Thanks to real-time forest footage transmitted by a smart management platform, the area in Tianquan County, Sichuan, has launched a live online wildlife streaming service. It once attracted more than 500,000 viewers in a single hour -- a record that has both raised public conservation awareness and driven local eco-tourism, as well as educational and research activities.
In addition to figuring out "where" the wild giant pandas are, technology is now being used to more accurately identify individual animals.
Traditional field monitoring methods relied primarily on analyzing the length and characteristics of bamboo stem fragments found in giant panda feces for rough identification. In the Tangjiahe area of the Giant Panda National Park in Sichuan, technological upgrades now allow staff to collect fresh wild panda feces and perform DNA testing to identify individuals. This enables them to accurately obtain core data, including sex, age, health status, and activity patterns.
Building on this foundation, the area saw another technological breakthrough in panda conservation last year. "We have fully rolled out AI-powered 'panda face recognition' technology," said Xiao Mei, an official with the administrative office of Tangjiahe National Nature Reserve.
Researchers trained the AI model on a massive dataset of field images, developing a customized intelligent recognition system for giant pandas. Once the system is fed with monitoring data from all infrared cameras in the area, it can automatically distinguish individual pandas in the footage.
The Tangjiahe area now operates a dual monitoring system that combines fecal DNA analysis with AI-powered panda face recognition. In 2025, the system detected a total of 44 individual giant pandas. It has clarified the size of the wild panda population, individual traits, and activity ranges in the area, providing scientific data to support wildlife conservation and regional ecosystem restoration.
Eco-environmental research experts from the Chinese Academy of Sciences noted that digital tools are shifting giant panda conservation from an "experience-driven" to a "data-driven" approach. With AI image recognition technology, the efficiency of monitoring wild panda populations can be increased by more than 40 percent.
Beyond facial recognition for wild pandas, AI is also poised for further application in monitoring the behavior of captive giant pandas. According to a paper published in February in the international journal Applied Sciences, the DeepPanda model developed by a Chinese research team can automatically identify four basic panda behaviors: eating, walking, resting, and drinking. Under experimental conditions, the model achieved an identification accuracy of 98.8 percent.
China not only leverages technology to support giant panda conservation but is also willing to share its achievements with the world and lead international cooperation.
Last November, a shared platform for giant panda ecological big data was launched at the Global Panda Partners Conference 2025 in Chengdu. This platform aggregates real-time monitoring data from 4,194.26 square kilometers of core zones within the Giant Panda National Park and provides 12 types of data interfaces, including vegetation coverage and population activity, to research institutions worldwide. Enditem
京公网安备 11010802027341号