British scientists have unveiled plans to create a digital library of all life on Earth. They say that the Digital Automated Identification System (Daisy), which harnesses the latest advances in artificial intelligence and computer vision, will have an enormous impact on research into biodiversity and evolution.
Daisy will also give amateur naturalists unprecedented access to the world's taxonomic expertise: Send Daisy a camera-phone picture of a plant or animal and, within seconds, you will get detailed information about what you are looking at.
At the Natural History Museum in London, Norman MacLeod, keeper of palaeontology, has spent several years developing the new technology.
He said that Daisy will make the identification of plants and animals more objective and directly comparable. "Right now, taxonomy is as much of an art form as it is a science,'' MacLeod said. He would present his vision for Daisy to an international meeting of taxonomists at the museum yesterday.
Taxonomists normally identify specimens through a painstaking process in which the features of an unknown plant or animal are compared with identified specimens in the museum's collections. If it is sufficiently different, the unknown specimen is confirmed as a new species.
However, there is plenty of room for error -- the museum's collection might be incomplete or the person making the identification could make a mistake.
If scientists did not have to make routine identifications and teach others how to do it, MacLeod argues, they could get on with the business of learning more about biodiversity and evolution.
"Say you saw a butterfly, you might take a digital image of it, connect up to the World Wide Web and access a Daisy Internet portal," MacLeod said. "The portal would accept the picture and farm it out to the servers in individual institutions, such as the Natural History Museum."
Using pattern-recognition software, Daisy would try to match the picture with images in its archives. "The portal would route the answer back as a Web page that had the confidence level of the identification and the institution that made the identification,'' MacLeod said.
Daisy can also identify sounds and scans of DNA barcodes.
"New developments in artificial intelligence and computer algorithms have taken neural nets to where they act more like human intelligence," he said. "When we see something new, we don't have to re-compute our understanding of everything else we've ever seen, we just add it to the mix. That's pretty much what we're doing with Daisy."
"Now there's a tool that we can use to justify making the investment in getting these collections of images together and building the software structures that are necessary to make the neural net able to access the images, then there's a reason to do it," he said.
There is also a role for amateur naturalists in improving the library. "One of the neat things about Daisy is that, if you submit an image and it's identified with a high level of certainty, that can then be added to the library of images, which makes Daisy more powerful," MacLeod said.
"That information can keep growing," he said.
"The more people that use [the system], the better it gets," he said.
He said that the first images and sounds have already been used to test that everything works. But filling Daisy with data from all the museums will take several years. The Natural History Museum has 70 million specimens that would need to be entered into the database.