Seeing Through An Animal’s Eyes
Computer Vision and Facial Recognition have gone from rare and difficult to increasingly available thanks to Artificial Intelligence (AI). But, we did not know how our brain sees – until now. Imagine seeing through another animal’s eyes.
Vision – an amazing capability of how we perceive and interact with our physical world. How the mechanics work in biological systems, like us, has been elusive. Nonetheless, we have been trying to get our computers to see objects and even recognize faces. We’ve been banging away at this challenge for decades but in the last year, breakthroughs in machine learning and artificial intelligence (AI) have enabled us to make great strides.
Things are progressing so fast that not only can you use facial recognition software to sort your photos when you store them but NEST (the Google company that makes smart thermostats, etc.) just announced that you can buy their home security camera with facial recognition to identify who is at your premise. Machine vision driven by AI enables computers and robots to perform all sorts of activities that once only humans could do.
This is definitely pretty head spinning stuff. But, while impressive it’s nothing compared to what Mother Nature can do. We get there by brute-forcing the solution using tons of computing power (brought to you inexpensively by the cloud). What is really elegant is how biology pulls off the same trick in a small space, with neurons and just a little bit of power.
Well, we now have an answer. Ironically, just about the same time NEST announced it was bringing computer vision to your doorstep, researchers from Caltech announced that they had cracked the brain’s facial recognition code. Imagine seeing with another animal’s eyes. Through a clever and sophisticated approach, they could actually “read” what a macaque’s brain saw. Sounds like science fiction. But see (no pun intended) for yourself.
Here are pictures of the faces they presented to the monkey test subjects compared to what they predicted the face would look like from reading the brain:
Even more astonishing, the scientists only needed to take readings from 205 brain cells. (Just to give you a sense of scale: there are estimated to be 100 Billion brain cells riding up there behind your eyes.) Talk about an incredibly efficient system to recognize faces.
Where do we go from here? One thought is: now that we know how we recognize faces perhaps we can use that to vastly improve our computer facial recognition algorithms and devices. Here’s another albeit a little more out there: if we can do this with one of our ape relatives, how long before we will be able to do the same with one of us. Imagine seeing through another person’s eyes!
By John Pientka