The conventional census methods, how much ever technologically advanced, are time taking, tedious and error-prone

Biometrics is not unique to humans but spans the entire living world. Wildlife conservation takes advantage of this fact to identify and keep count of wildlife, particularly endangered species. The Alan Turing Institute, a center in the UK for data science and artificial intelligence went a step ahead with wildlife research, employing Artificial Intelligence and machine learning technologies to build a model for Tiger conservation. The research led by Debbie Banks living in a north-eastern town in Scotland aims to collect as much data about tiger skins as possible to feed it to a machine learning algorithm. So far, she has collected thousands of photographs which include dead animal parts, rugs, and taxidermy specimens. She is an environmental activist working for EIA (Environmental Investigation Agency), a London-based charity organization. The project is pivoted on the fact that no two tigers share a similar pattern of stripes. Once the database of a particular tiger skin is uploaded it becomes easy to find out from where it comes. "A tiger's stripes are as unique as human fingerprints. We can use the images to cross-reference against images of captive tigers that might have been farmed ", told Banks to AFP.

AI Vs conventional methods:

The Tiger census goes back to 1932 when the world's first Tiger census was carried out in the Palamu Tiger reserve, essentially using pugmarks to identify and count tigers. In course of time, the methods took technology's help to take close shots of live tigers roaming around strategic points like water bodies in the forest. DNA analysis of dead body parts, whisker analysis, poop analysis, and satellite telemetry helped scientists to large extent in building a census of tigers successfully. The traditional methods how much ever advanced technologically were time taking and tedious and at times error-prone. In 2018 artificial intelligence made its foray into animal conservation thereby making it an algorithm-based technique. But then, the technique was limited only to training the machine learning algorithm to differentiate between a Tiger and other wild cats. The research was based on teaching AI to understand how a tiger looks by also learning what a tiger doesn't look like. And then the task of identifying individual tigers within species was undertaken. The experiment carried out in the Tadoba forest was successful at 96% which turns out to be an impressive number but not close enough to depend on the model. A Tiger ID app was used to feed thousands of images collected from tiger lovers to generate signatures for more than 3000 tigers. The new machine learning tool developed by Turing Institute is aimed at cross-referencing tiger identity digitally through images, leveraging its unique stripe pattern. "At the moment we are doing that manually, looking at the individual stripe patterns of each new image that we get and cross-referencing it against the ones we have in our database", said Banks.

AI models as a key to sustainable living

While researchers are struggling to find ways to nature conservation and species identification, which to a large extent remains a human-led process limited to robots and drones, artificial intelligence has taken the game to an entirely different level. AI algorithms are as intelligent as humans to understand the random behavior of wild animals. If trained to perfection with the data required – a major challenge for any machine learning model – the very versatile artificial intelligence can pave the way for sustainable living. To make their model error-free scientists at Turing Institute have asked for photographs from photographers, researchers, and academics who possess images of tigers with obvious stripe patterns. Who knows the one lazy holiday picture taken at a zoo might expose a cross-country poaching scandal?