New Flower Recognition API: be your own botanist

If you’ve ever struggled to identify a flower or plant we’ve got the answer. We’re excited to announce our new improved flower recognition technology. Simply point the Blippar app (on iOS or Android) at a flower to give it a try.

Our technology uses deep learning to recognise over 400 flower species and genera. We’ve trained our platform based on the most common flowers found in the wild and rarer species found in flower shops both in the UK and USA. The technology is available now as an API for businesses to licence and anyone can test it using the Blippar app.

We used a broad dataset of more than 200,000 images and advanced deep learning to ensure our flower recognition software can distinguish between even minute differences in flower species -- matching the knowledge of an expert botanist. The technology reaches a classification accuracy of 91.65%, making it the most accurate flower recognition software on the market and the one with the lowest number of false positives.

In the images below, you can see how the technology can distinguish flower species even when they differ by only minor details. For example the difference between the very similar Alcea and Malva flowers, or between Jasminum Nudiflorum and Forsythia.

Our technology can also distinguish between different species of the same family, for example the various types of orchids below.

The API currently recognizes the most prominent flower in an image, with a few more advanced features available on demand. These include recognizing multiple flowers in an image, as well as localising their position within the image.

Notoriously difficult to describe using text-based search, the development of visual search for flowers means that budding gardeners, florists, visitors to botanical gardens or anyone interested in flowers will have their curiosity satisfied quickly and easily. Especially if you've ever needed to identify a flower by a picture.

This is the latest step in our aim to catalogue and index the entire physical world, and to create a visual search tool that enables smartphones to see the world like humans do.

As the UK government seeks to establish the country as a leading hub for the development of artificial intelligence, the advances made by Blippar and other computer vision pioneers are critical to the creation of a thriving AI ecosystem.

Ambarish Mitra, CEO & Co-Founder of Blippar, commented: “This launch is another stride forward in our mission to break down the barriers between the physical and the digital world. One of the key utilitarian use cases of computer vision is flower recognition and that of nature in general. Words are not enough to describe many of the things we see in nature and this is where this technology answers a gap. Our flower recognition technology now matches the accuracy and knowledge of an experienced botanist, allowing users to access information on the plants and flowers surrounding them in their day-to-day lives. This is another testament to our pioneering experts in computer vision and an indication of the future impact augmented reality is set to have on our everyday experiences.”

If you are interested in integrating our flower recognition technology into your app or website, get in touch. Or if you ever need to identify a flower by a picture or in real life, try using the Blippar app!

For more information, read our official press release.