Slyce and the Updates on Visual Search Technology

Pinterest and Shoes.com are testing out a deep learning software that tremendously improves the shopping experience with image recognition tools. With both companies, users can search items by selecting parts of an image for more enhancement of the features.

Even though the Internet has changed a lot in 20 years, the text search boxes haven’t. Image recognition software are used to challenge that, and both companies are using a technique called deep learning.

Deep learning has recently enabled software to match people on some distinctions for image recognition. The new visual search tool for Pinterest allows users to draw a box around something on an image on the service to find similar, visual items from an index of over a billion. As a feature, testing example, drawing around a coffee maker in a kitchen photo turned up similar ones, including closed up photos of the same model. Pinterest also introduced over the past summer, a buy button attached feature for some items with the visual search. And now, the image search function is presented to all of the company’s website users and mobile apps. Pinterest’s system has learned to understand images by the attached text from people being drawn to photos shared on the service.

Kevin Jing, who is the head of Pinterest’s visual search, stated that the visual search has better chances in becoming indispensable, in which image representation that comes from deep learning is a lot more accurate. He further stated there has been much more improvement, even though the visual search tools are not perfect. But when a lot of people would use them, it will become clear whether or not the technology has improved enough to alter how people interact with online services.

In the past, companies tried to use image search technology to make the discovery and shopping of products easier. For example, Amazon used an app to look for products that are snapped in a photo with the Fire smartphone last year, but it was unsuccessful. In 2010, Google purchased Like.com, and this website launched a shopping comparison site used to find products visually similar to the product selected. It even allowed users to highlight important, image details to guide the selections. And as a result, Google’s visual search tool allows for visually similar products, but there is no capability to highlight the details.

Shoes.com is testing a different technique to visual search that is also from deep learning. Being the first to use the image processing technology, this technology is developed by the artificial intelligence, startup company called Sentient, who has gained $143 million from investors.

Shoes.com are beginning to test Sentient’s technology in the women’s boots area of the store. Users must click on “visual filter”, and then a grid of 12 images that represent the most distinct, style clusters from a catalogue of around 7,000 boots are shown. The user must then select the closest one he/she is looking for, and the software would utilize the visual characteristics of the selection chosen to refresh the grid in showing 11 more that are similar. When repeating the process a few times, the selection with the very particular characteristics of boots are shown.

Another great company for retailers to use for visual searching of products is Slyce. With their experienced and leadership team in this field of technology, they are significantly making a great shopping experience for customers and retailers.

For one thing, Slyce has a universal scanner that is a great solution used for the most accurate, image recognition devices available today. The visual search technology incorporates the top features of several of their competitors in one product solution. In solutions for e-tailers and retailers, the use of image recognition technology are available in both desktop and mobile devices, in which activation of visual product recognition of existing product images, or the simple snapping of a photo is granted.

Leave a Reply

Your email address will not be published. Required fields are marked *