As the internet matures and search engines getting better in what they do, people have come to expect them to automatically understand their intentions: people want answers from what they ask on a single search box.
But that case is difficult. The web is large and the number of websites that are added each day are still increasing. What's more, computers that are now smart, should always be smarter in the future. So enough is never enough.
As a search engine of the web, Bing's popularity is second if compared to Google. But doesn't meant that it is shy from introducing new things for better experience.
Over the years, the search engine by Microsoft updates and upgrades its systems just to be better. And on June 1st, 2017, it's announcing one of the good things that can make web searches a lot easier.
Available as part of the Bing's existing image search tools, it's called 'Bing Visual Search'. What it is, is an expansion of its image search toolset that lets users search for specific items shown within a larger image.
In an example below, users want to search for a bowl that is shown inside a larger image.
In the Detail View, users are presented with a magnifying glass symbol located in the top left of an image. This is the visual search button.
Clicking on the button will open a visual search box for the said image. There, users can click and drag a box to adjust it to cover just the object (bowl) they want. Users can also draw a line around the object if it's more convenient.
Each time users adjust the position or the size of the visual search box, Bing will run a visual search using the selected portion of the image as the query.
And if Bing detects users' shopping intentions, the search engine will also run a product search to find matching product in addition to its regular image search.
To make this feature happen, Bing needs to process images like how it processes text.
The differences is that Bing runs the image to its Image Processing Service to perform object detection, extraction, recognition, etc.. Then it passes the data to Text Query Inference where it generates the best text query to represent the input image. Subsequently, Bing also runs a Scenario Triggering to identify the context of the image.
Bing Visual search is available on PC or mobile devices by visiting Bing.com or its mobile app, and should work with existing internet images as well as new user-taken photos. Bing is also making visual search available to developers via its image search APIs.
"Visual search is in its infancy, and we are aware of cases where there is still room for improvement," said Bing in its blog post announcing the feature.
Previously, competitors are already having similar features using similar technology. For example, Google Search has been using product metadata to identify items seen in a photo. Pinterest has a Shop to Look feature where it turns items inside an image into something users can buy. Instagram is also testing similar feature.