New research has demonstrated for the first time that artificial intelligence (AI) can recognise individual birds in the wild. It is the first successful attempt at performing individual recognition on small birds and uses technology which is similar to software that remembers faces on social media.

The research, published in the British Ecological Society journal Methods in Ecology and Evolution, was a collaborative effort by researchers at the FitzPatrick Institute of African Ornithology, based at the University of Cape Town (UCT) in South Africa, the French National Centre for Scientific Research, the Research Centre in Biodiversity and Genetic Resources and other institutes from Portugal, and the Max Planck Society from Germany.

Dr André Ferreira at the Centre for Functional and Evolutionary Ecology, France, and lead author of the study said: -

“We show that computers can consistently recognise dozens of individual birds, even though we cannot ourselves tell these individuals apart. In doing so, our study provides the means of overcoming one of the greatest limitations in the study of wild birds – reliably recognising individuals.”

The research was done on three species of birds including the wild great tits, sociable weavers and captive zebra finches, some of the most studied birds in behavioural ecology.

The international team trained the AI for individual re‐identification (i.e. machine recognition of a previously known sets of individuals) where data collection was done both in the wild and in captivity.

The birds carried a PIT (Passive Integrated Transponder) tag, an internal electronic microchip with unique barcodes, similar to those used in pets dogs and cats. When the antenna on the feeders detected one of these tags, it would then trigger the cameras to take photos of the birds. Thousands of these labelled images were then fed to a computer, which used a deep learning algorithm known as Convolutional Neural Networks (CNN’s) to train and test AI models. CNN’s are optima for solving image classification problems. The AI was able to accurately identify over 90% of the wild great tits, 92.4% of the sociable weavers and 87% of the captive zebra finches.

The ability for AI to recognise birds individually is critical in order to understand their needs, population migration routes and increase species protection. An improved understanding of these areas will necessarily help implement better conservation efforts and assist in curbing existing issues, such as climate change, agriculture expansions, logging and farming and deforestation.

However, the authors of the study cautioned a limitation of the AI model, in that it is only able to re-identify individuals it has been shown before. If new birds join the study population, the computer will not be able to identify them. A further concern for the researches was that the birds may also experience temporal changes in appearances such as molting.

The research team is nevertheless confident that these limitations can be overcome with a large enough datasets, which they are currently collecting.

Courtesy of Amelia Wang

Main resource: Metro UK