Computers are starting to develop their own senses – and the experts are starting with the sense of sight. This all began with the simple face recognition feature in cameras, but the tech world has other things in store for the public.
So far, the focus is on visual search – and companies continue to race in developing this ground-breaking function.
Naming the developments
One of the main developments in visual search is the Google Goggles. This is one of the pioneering ventures toward visual search. Through this, you can easily snap a photo of something and the search engine will do its best to find information regarding the photograph’s contents.
Amazon is also experimenting with visual search. As a matter of fact, the retail giant included visual search as a key feature in one of its smartphones. Many other smaller image-recognition start-ups continue to research on this matter as well.
Explaining the technology
What makes this seemingly futuristic technology possible is another discussion. After all, the science behind visual search can get complicated. Images need more processing power than plain text, which means the technology needs more complex search algorithms.
One algorithm companies focus on today is object recognition. There are two notable techniques used in this: edge detection and scale invariant feature transform.
Edge detection depends on the lighting changes in a particular image. Through edge detection, the computer gets an idea about the object’s shape and colour, thus making it possible to construct a search algorithm out of it.
The second technique directs the computer to extract a set of reference images from a certain object. The algorithm, then, uses these images to tell whether or not the search has a match in the database. This makes it possible to look for video files like, say, AE templates or stock footage through the use of still images.
Object recognition, however, is just the tip of the iceberg. So far, companies continue to explore the role of GPS and geo-tagging in terms of filtering the results for a particular image search. This goes on top of the usual side-by-side run of text search and image search.
There’s a high chance that visual search would become central in helping search engines and e-commerce sites provide more relevant results to their users. Sure, the technology may still be in its infancy; but with the pace of development in the tech world, who knows what the next decade or so holds?