New AI can distinguish birds by their appearance

The British Environmental Society has developed artificial intelligence that can be used to train computers to recognize individual birds by their appearance. The study is published in the British Ecological Society journal Methods in Ecology and Evolution.

Computers can sequentially recognize dozens of individual birds, even if we ourselves cannot distinguish between these individuals. Thus, our research provides an opportunity to overcome one of the biggest limitations in the study of wild birds – the identification of individuals.

André Ferreira, Doctor of the Center for Functional and Evolutionary Ecology (CEFE) in France

The study involved scientists from institutions in France, Germany, Portugal, and South Africa. They describe the process of using artificial intelligence to individually identify birds. This involves collecting thousands of labeled bird images and then using that data to train and test artificial intelligence models. This study represents the first successful attempt at AI-assisted bird identification.

The researchers trained artificial intelligence models to recognize images of individual birds in wild populations. After training, the AI ​​models were tested on images of different individuals that they had not seen before. As a result, the algorithm named all individuals with an accuracy of 90% for wild species.

In animal behavior studies, individual species identification is one of the most expensive and time-consuming behavioral and population-limiting factors that researchers can study. Modern identification methods, such as attaching colored stripes to the legs of birds, can also be stressful for animals.

These problems can be solved using artificial intelligence models. The ability to distinguish individual animals from each other is important during long-term monitoring of populations.

However, the appearance of individual birds can change over time, for example, due to molting, and it is not known how this will affect the performance of the artificial intelligence model. Images of the same bird taken several months later can be mistakenly identified as different individuals.

Tags: ,