Birdsnap app analyses your photos to tell you what bird you've snapped
comments
It's not uncommon to see a pretty bird flit by when you are enjoying a walk or a picnic - even if you live in a built-up area.
But if your knowledge of flying fauna isn't up to Bill Oddie's standards, it can be difficult to work out which creature you have spotted.
But now an app, developed by computer scientists, that can identify a bird from a single photo.
Watch the birdie: Computer scientists have developed an app that can identify birds from a photo using computer vision and machine learning techniques. Users can also browse though species (pictured)
A team of researchers led by Professor Peter Belhumeur at Columbia Engineering, New York City, have used computer vision and machine learning techniques to create Birdsnap.
The free iPhone app claims to be an electronic field guide featuring 500 of the most common North American bird species.
It lets users identify bird species by uploading photos and there is also a website, which includes 50,000 images as well as bird calls for each species.
Beginners and experts can sort through species alphabetically, by their relationship in the Tree of Life and by the frequency with which they are sighted at a particular place and season.
Picky: The app detects parts of a bird so that it can examine the visual similarity of its comparable parts. It automatically discovers visually similar species and makes visual suggestions for how they can be distinguished (pictured left) and information about the different birds can be scrolled through (right)
It automatically discovers visually similar species and makes visual suggestions for how they can be distinguished.
THE BIRDSNAP APP
Birdsnap app was developed by computer scientists.
It uses computer vision and machine learning techniques to identify species of birds from photos.
'Electronic field guide' features 500 of the most common North American species.
It is designed for iPhones and is free from the App Store.
Users can upload photos and the app will identify birds in them.
They can also use a website to browse through the species in different ways and listen to different bird calls.
In the future the app could offer a Shazam-like feature to recognise bird song.
The technology could also be added to binoculars to identify and tag species within the field of view.
The app and website were developed with the University of Maryland and made their debut at the EEE Conference on Computer Vision and Pattern Recognition in Columbus, Ohio.
'Our goal is to use computer vision and artificial intelligence to create a digital field guide that will help people learn to recognise birds,' Professor Belhumeur said.
He has previously launched Leafsnap - a similar electronic field guide for trees.
'We've been able to take an incredible collection of data - thousands of photos of birds - and use technology to organise the data in a useful and fun way.'
He and fellow computer scientist David Jacobs of the University of Maryland, realised that many of the techniques they have developed for face recognition, in work spanning more than a decade, could also be applied to automatic species identification.
Face recognition algorithms rely on methods that find correspondences between comparable parts of different faces, so that, for example, a nose is compared to a nose and an eye to an eye.
Birdsnap works the same way by detecting the parts of a bird so that it can examine the visual similarity of its comparable parts. Each species is labelled through the location of 17 parts.
It automatically discovers visually similar species and makes visual suggestions for how they can be distinguished.
Catalogue of calls and clucks: A website (pictured) includes 50,000 images plus bird calls for each species. Beginners and experts can sort through species alphabetically, by their relationship in the Tree of Life, and by the frequency with which they are sighted at a particular place and season
'Categorisation is one of the fundamental problems of computer vision,' said Thomas Berg, a Columbia Engineering computer science PhD candidate.
'Recently, there's been a lot of progress in fine-grained visual categorisation, the recognition of - and distinguishing between - categories that look very similar.
'What's really exciting about Birdsnap is that not only does it do well at identifying species, but it can also identify which parts of the bird the algorithm uses to identify each species.
'Birdsnap then automatically annotates images of the bird to show these distinctive parts - birders call them "field marks" - so the user can learn what to look for.'
The team is also using the fact that smartphone cameras embed the date and location in their images to improve classification accuracy. They have designed a system that can pinpoint which birds are arriving, departing, or migrating.
The scientists hope to add the ability to recognise bird songs in the future as well as creating 'smart' binoculars that use this artificial intelligence technology to identify and tag species within the field of view.
I spy: The scientists hope to add the ability to recognise bird songs in the future as well as creating 'smart' binoculars that use this artificial intelligence technology to identify and tag species within the field of view. A girl using standard binoculars is pictured
Put the internet to work for you.
0 comments:
Post a Comment