The new image search interface was launched last year along with Google Voice Search for Android, and was very well received. It gave people the option to search for a topic based on a picture they had, whether to find a better quality version of the image, finding source of the picture or just looking up some information related to it.
The interface started off well enough but didn’t deal well with image distortion. Now, due to monthly algorithm updates the interface has become faster and a lot more intuitive. Now for the first time since its release, they have released a expansive format update.
More Intelligent Best Guesses
Whereas before the above picture would have likely resulted in a guess like ‘flower’ or petunia., now the interface will try to specifically identify the exact contents of the picture down to species, sometimes.
‘Best guesses’ were often not returned with certain images or were quite inaccurate but the new update will focus on a stronger result accuracy with all their image queries.
Broader Set of Results
With Google’s improved ability to recognise and define images, they can now integrate the search functionality of their other vertical search engines.
When a topic or definition comes up as related to the image you will have the option to search these results along with the usual web results. So if you search for an image that comes up as ‘Venice’, for instance, you will be able to see the latest news on Venice, other images related to the topic or even see satellite images of it on Google Maps.
The emphasis of this move seems to be on the freshness of content, getting the most up to date information relating to that picture.
Knowledge Graph Integration
Another relatively new feature on Google. With the improved image recognition, it seemed logical for Google to include this topic summary interface.
If an image returns a well known topic or person as its best guess, then along with the other results, it will give a concise overview of this best guess and suggest some related topic, books and videos for example. This section is placed on the right hand side of the results.