"My eyes are the ruler"! How can artificial intelligence help us understand the "family background" of birds?

Biodiversity includes animals, plants and microorganisms, among which birds are one of the categories that attract the most public attention. Because they are sensitive to habitat factors and their changes, they are often used as a "barometer" and "touchstone" of regional ecological quality. Therefore, bird monitoring has become one of the regular tasks of relevant units.

File photo: Staff members are investigating and photographing birds

Initially, bird monitoring mostly adopted the method of "manual + observation equipment", mainly through inspectors walking through every corner of the protected area, looking at them one by one, and counting them one by one, which was time-consuming and laborious, and the detection efficiency and accuracy were very low. There were some dangerous tidal flats and swamps that the inspectors could not enter, which became statistical blind spots. With the rapid development of artificial intelligence technology, it has become possible to assist bird diversity monitoring with AI-related technologies. After the bird intelligent identification system was launched, staff can carry out monitoring work indoors, video surveillance has a wider coverage, and monitoring efficiency, accuracy and feasibility have been greatly improved. As sample data and sample size gradually increase, the recognition accuracy is also getting higher and higher. For example, now the monitoring accuracy has reached more than 90%. This not only provides innovative means for bird diversity surveys and dynamic monitoring, but also brings "smart changes" to wildlife monitoring.

How to use AI for a "bird census"? First, AI bird investigators need to install an "AI bird intelligent identification monitoring system", which consists of hardware monitoring equipment and intelligent identification software. Secondly, the bird AI identification server has the functions of real-time bird monitoring and efficient analysis and processing. The bird monitoring and management system has integrated functional applications such as video access, monitoring and control, image management, bird list, and statistical analysis. The core of the bird intelligent monitoring and identification system is the bird intelligent identification algorithm, which mainly includes bird identification algorithms based on images and videos, and bird identification algorithms based on sound. Bird identification algorithms based on images and videos are more suitable for relatively open natural environments such as wetlands or water surfaces. For example, the Oriental White Stork, the system automatically identifies its obvious identification features such as red skin around the eyes and black beak, and the identification features of the white spoonbill's black and flat beak, so as to make identification judgments. In environments such as mountains and forests with severe obstruction, the sound-based bird identification algorithm is more effective.

AI has been applied to bird diversity surveys and monitoring and migratory bird detection in many protected areas and wetland parks. Artificial intelligence has been applied to bird monitoring in the Yellow River Delta National Nature Reserve. Since its operation in 2022, the system has provided rich video monitoring data, providing strong data support for the managers of the reserve to conduct real-time monitoring and view historical records.

In the migratory bird season, artificial intelligence can easily be used to count the spectacular waves of birds. In September 2022, the Kunming Dianchi Plateau Lake Research Institute deployed a bird intelligent monitoring and identification system at Dabokou. When the red-headed gulls arrived, the system could monitor more than 5,000 red-headed gulls every day, providing a large amount of observation data for the managers of the reserve.

In the future, emerging scientific methods and tools will be applied to bird monitoring, allowing bird monitoring to develop in the direction of high precision, dynamism and intelligence, significantly improving the efficiency and identification accuracy of bird monitoring, promoting scientific research and protection of birds and their habitats, and providing corresponding technical support for biodiversity conservation and popular science publicity, contributing to our better understanding and protection of birds.

Audit expert: Zheng Chaochao, Senior Engineer, China Electronics Engineering Design Institute

China Association for Science and Technology Department of Science Popularization

Xinhuanet

Co-production