A "sixth mass extinction" is underway, as per the latest study titled “The Sixth Mass Extinction: fact, fiction or speculation?” It further states that the Earth could have already lost about 7.5 and 13 per cent of its total species. The trend is signalled by widespread vertebrate extinctions that will have negative cascading consequences on the functioning of the ecosystem and services vital to sustaining civilisation. 

There is a growing recognition among wildlife researchers to ensure fast assessment, response, and review as it is required to understand and combat the continuous decline. And this is where AI can come to the rescue.

Dr Tanya Berger-Wolf and her colleagues are pioneering imageomics - a new field of study, as per an interview given to The Verge. Imageomics employs machine learning to extract biological data from photographs and videos of living organisms, as the name implies. Berger-Wolf and her colleagues have lately begun working with Indian leopard researchers to compare spot patterns of mothers and children using algorithms.

AI can aid conservation research

The team is currently working on an open-source platform called Wildbook that aids wildlife researchers in collecting and analysing pictures. They are presently concentrating on generative AI approaches. As per the team, Wildbook provides a technical foundation (database, APIs, computer vision, etc.) for wildlife research projects to:

• track individual animals in a wildlife population using natural markings, genetic identifiers, or vocalizations
• collect biological samples from a wildlife population and perform genetic and/or chemical analyses (e.g., stable isotope measurements, haplotype determination, etc.)
• engage citizen scientists and\or using social media to collect sighting information
• build a collaborative, distributed research network for a migratory and/or global species
• develop a new animal biometrics solution (e.g., pattern matching from photos) for one or more species
• collect behavioural and/or social data for a wildlife study population

Moreover, the platform scans millions of crowdsourced wildlife photographs at scale using computer vision and deep learning algorithms. The public can track the travels of their favourite animals using Wildbook, which can identify species as well as individual animals. Scientists utilise the aggregated data to help them make conservation decisions. Microsoft is assisting them by hosting Wildbook on Azure and making them open-source algorithms developed by Wild Me available as APIs.

Come to India, the National Tiger Conservation Authority (NTCA) and Wildlife Institute of India (WII) have, in collaboration with IIIT-Delhi, conducted a nationwide assessment of the country's population and habitat of tigers (Panthera Tigris) using their "AI-based wildlife monitoring." This has been termed the most comprehensive census to date in terms of both the resources and the data amassed. This project also gets featured in the GUINNESS BOOK OF WORLD RECORDS for the "Largest Camera Trap Wildlife Survey".

Similarly, researchers can get hold of millions of images and videos of animals in their natural habitats without the interference of humans or captivity using cameras disguised in wildlife habitats. These "camera traps" are essential for understanding our wild ecosystems, as they provide information on behaviour, migration, and population dynamics, among other things. Take Zamba Clouds, it is a freely available web application that does camera trap video processing work utilising state-of-the-art computer vision and artificial intelligence, freeing up more time for humans to focus on understanding the information and using the results.

Conclusion

AI has the potential to automate and hence accelerate data annotation, allowing for real-time engagement and decision making. Deep Neural Networks, Convolutional Neural Networks, and other machine learning methods can be used to recognise useful animal behaviour in audio and visual data. It is important to note that researchers can focus on deeper analysis that can lead to actionable change if this baseline information is reliable.