Why image recognition?
There is an ever increasing amount of digital visual information being produced. For instance, there are over 800 million mobile phones with integrated cameras. These cameras are increasingly used as a tool in order to remember or share things by just snapping pictures. Have a look at your digital photo collection today, and imagine how large it is gonna be in 30 years.
And did you know, that each day 250 million photos are uploaded to Facebook? So, there is huge amount of digital images being produced, but organizing and searching them by their contents is not possible without image recognition.
The current state-of-the-art in Computer Vision allows for reliable recognition of Print and Paintings (including Newspapers, Magazines, Catalogues, Billboards, Postcards, Brochures, Flyers, Calendars, Christmas Cards, etc.), most Products (including CDs, DVDs, Books, Games, Consumer goods with label or printed packaging, Wine labels, etc.), and Places (including Landmark buildings, Statues, Houses, Scenery, etc.).
All of the above can be recognized in 1 second or less.
Things that don't work well or not at all:
- Object classes (e.g. plants, food, animals, ...)
- Shiny objects (cars, TV's, cameras, ...)
- X-Ray and other medical imagery
"Vivino - as the world's largest wine recognition app - highly relies on a robust and fast recognition system, even under bad conditions. We drink wine in bad light and the angles of the images vary a lot from image to image. The scalability of kooaba's image recognition technology is amazing. We already have over 700,000 images in the database and it matches in less than a second."
- Heini Zachariassen, CEO Vivino