Home / Blog / Using AI to detect animals eyes

Using AI to detect animals eyes

Different kinds of eyes in the animal kingdom. Our brain promptly recognise animal eyes (well, not all of them, like the viper pit doesn’t look an eye to us…) but for a machine, a digital camera, recognising animals eyes is a challenge. Image credit: MorgansList

Nature has evolved an impressive variety of eyes, not just in shape but also in characteristics. You may want to take a look at this wonderful article, from where I extracted the photos of a variety of eyes, to “see” what animals can “see”. I found it an amazing and thought provoking piece of information (did you know that the mantis shrimp has the most sophisticated eye so far found on this planet? I didn’t).

Our brains have acquired the capability to spot an animal eye … in a blink of an eye (!), yet there is a huge diversity of eyes in the animal kingdom. Point and shoot digital cameras, and more recently reflex digital cameras have acquired the capability to recognise human eyes to set the focus on them. This is important since our brain is trained to look at the eyes as the dominant feature of an image and this holds true when we look at a picture. If the eye is sharp on focus the whole photo looks better. Hence the importance in photography to get the eye(s) sharp.

Digital cameras are using quite a bit of real time processing to search for an eye in the frame and once they find it direct the focusing to have it as sharp as possible. Now this is tricky because they need to detect an eye in a frame that is not in focus and because there are different shapes of eyes and eyelids, different colours and the eye can be in different positions. Increasing processing / storage capabilities and artificial intelligence have allowed this feature in modern digital cameras.

Now Sony has announced a software update to its a7R III and a7 III digital cameras enabling automatic animals eye focus. This is a significant improvement that for sure will delight wildlife photographers (and be a welcome feature for pets lovers). In wildlife photography getting a sharp focus on the eye is essential, and yet it is usually quite complicated since wild animals seem to have their own mind in terms of acting like a model (they don’t). Having the camera taking care of focus can let you … focus on framing and on the artistic side of photography, rather than on the technical side.

As Sony is pointing out, the variety of shapes and colours of animals eyes is staggering and training the camera software to detect animals eye has been quite a feat (actually they are saying that over the coming months the software will keep learning and new releases can be expected to broaden the set of animal eye recognition).

Artificial intelligence, and in particular machine learning and computer vision, plays a key role and what is impressive is the speed achieved in eye recognition. Speed, of course, is of paramount importance in photography, even more so in wildlife photography. The camera software has to make decision in a split second on the presence of an eye in the frame and then direct the focusing mechanism to make that point as sharp as possible. The algorithm has been trained on hundred of thousands of images containing animal eyes and has learned to recognise them.

What seems straightforward to us is the result of million of years of evolutionary training. This is now condensed in a few months of software training. Amazing, isn’t it?

About Roberto Saracco

Roberto Saracco fell in love with technology and its implications long time ago. His background is in math and computer science. Until April 2017 he led the EIT Digital Italian Node and then was head of the Industrial Doctoral School of EIT Digital up to September 2018. Previously, up to December 2011 he was the Director of the Telecom Italia Future Centre in Venice, looking at the interplay of technology evolution, economics and society. At the turn of the century he led a World Bank-Infodev project to stimulate entrepreneurship in Latin America. He is a senior member of IEEE where he leads the Industry Advisory Board within the Future Directions Committee and co-chairs the Digital Reality Initiative. He teaches a Master course on Technology Forecasting and Market impact at the University of Trento. He has published over 100 papers in journals and magazines and 14 books.

Leave a Reply

Your email address will not be published. Required fields are marked *