Almost since the inception of the web companies have been trying, and mostly failing, to develop a visual search engine using image recognition. It is only within the last few years that visual search engines have come into their own and now is an exciting time for image recognition.
A visual search engine is one that matches an input image against the billions of images available on the web and returns a visual representation of images that match the input image. These searches can be done for any number of reasons. For example, someone might search for a product to see if there is a better price on another site or search for matching pictures of landmark to see where it is located.
There are a number of challenges in implementing a visual search engine. The first is that variations in the quality of the image including characteristics such as shadow, distance, time of day, and weather can make it hard to match against the same image taken with different characteristics. In order to better match images, visual search engines must combine features of an object in an image, such as texture, color and shape with image metadata, such as location tags and descriptors to find the best match.
A second, and even possibly even bigger, challenge is cataloging the billions of images available on the web. For example, as of June 2016 Instagram users have shared more than 40 billion photos. That is just one source, if you combine all of the photos on the web the number photos to be catalogued is in the hundreds of billions. Indexing and properly tagging all of these photos and then making them instantly retrievable by users is a daunting task as is keeping up with millions of new images being added every day.
That is why a lot of visual search engines have opted to focus on a specific category of image recognition. This specialization allows them to offer a more precise focus to a specific audience such as fashion or product searches.
Slyce is an image recognition company that has taken a broad approach to image search. It offers their clients a number of solutions that help their customers search on bar codes, QR codes and real world products to find matches all from the customer’s smart phone. Slyce offers a wide range of 1D, 2D, and 3D image searching opportunities.