Flickr Adds New Intelligent Recognition Search Tools
Flickr has implemented new intelligent search tools that recognise objects within users’ photos and sort them accordingly. The site is hoping that the new tools will encourage users to take advantage of the free 1TB of space available to them.
The intelligent recognition tools are able to analyse a picture to find out what’s in it as well as the colours the photo contains. Users are then able to search through pictures on Flickr to find what they’re looking for which can include an object, animal or place. The tools work on a user’s own photos as well as those that have been uploaded by others and made public.
In addition, the same technology is used by the ‘magic view’ window which categorises photos based on individual parts of the image. For example, Flickr can organise photos based on whether they have an animal in them and what type of animal it is. Photos can also be sorted according to whether they contain one person or a group of people and the technology will know whether the people are children or adults. In addition, photos can be sorted by style so that those that are abstract, vintage or black and white will be grouped together.
As well as sorting photos that have already been uploaded, Flickr is making it easier for users to upload their photos and have them automatically sorted. An automatic uploader appears whenever a memory card is inserted on Windows or Mac, which allows users to upload photos simply. These photos will be marked as private when they are uploaded so that the user can later make a decision about whether they should be shared.
Flickr is hoping that the new tools will encourage more users to take advantage of the 1TB of free online storage that the service started making available in 2013. At the time, the addition of a large amount of storage encouraged lots of new users to sign up, however they were then left with large collections of photos that had to be tagged before they could be sorted through. The new automatic tools make it easier for photos to be searched and sorted through as well as making it easier for people to use Flickr over other photo hosting sites. Large sets of photos can be selected which can then be transferred to an album, downloaded or kept for referencing.
Flickr uses ‘convolutional neural networks’ as the basis for its new tools, which are computers that act like the human brain. Even when the tool is wrong, it will make understandable human-like errors and will be able to learn from human feedback. For example it may confuse a domestic cat with a wild big cat and will be able to make less mistakes as users tell it what it has got wrong. In addition, feedback will allow it to learn about words and phrases, knowing that a phrase may refer to someone’s name rather than the objects it describes.
Under the previous search tools, if a user wanted to find a red bird, they would hope that either they or someone else had tagged appropriate photos correctly in order for a large number of images of red birds to appear. However, with Flickr’s new tools users can simply search for ‘bird’ and select the correct colour from the colour picker at the top of the page. They will then be presented with photos of red birds from their own photos or the public photo library, even when those photos haven’t been manually tagged.
The new updated tools are available for computers but are being launched on mobile devices too. While the intelligent search will be available for smartphones and devices, Flickr is hoping that the new tools will lure users away from other cloud based services such as Apple’s iCloud which offers less free space for images. Improved tools for mobile will also help to end reliance on tools such as IFTTT to upload photos to Flickr, or Dropbox to transfer them between devices. Users also won’t have to wade through years of unrelated images in order to find the set they‘re searching for.