Search and you will find

Search and you will find

IF YOU have searched for items on online shopping websites, chances are you have come across search results that do not match what you are looking for.

Singapore-based start-up ViSenze has devised a solution through its visual search and recognition technology that understands objects to find the best matching products.

“We saw how consumers struggled to describe what they were looking for using key words, yet immediately identify what they wanted upon seeing a photo of the product or something similar to it,” says Mr Oliver Tan, chief executive officer of ViSenze.

“If we could combine computer vision and machine learning algorithms into a tool that takes away the hassle of keyword guessing, we would quickly solve this inherently difficult real world problem,” he adds.

According to Mr Tan, ViSenze’s technology can detect and classify objects in images and videos — such as bags, shoes, jewellery and furniture — and extract visual attributes such as shape, pattern, colour granularity, as well as unique style and design features in the same way the human brain interprets these signals.

“ViSenze uses these visual signals to find exact or visually matching products from any large database without a need for keywords, within milliseconds,” he explains.

ViSenze, which was founded in 2012, currently processes tens of millions of image queries a day. Well known e-commerce companies such as Zalora, Flipkart, Lazada, Reebonz and Rakuten are using its technology.

Online fashion powerhouse Zalora, for instance, uses ViSenze’s visual recognition capabilities to power its recommendation feature, allowing consumers to find relevant and similar products from a plethora of brands.

Mr Tan attributes ViSenze’s success to its focus on solving real world problems.

“We weren’t interested in pushing technology to users. We were determined to identify the root problem shoppers faced in searching for products, and to provide a visual tool to help them.

“By constantly focusing our attention on this, we were able to make real headway,” he says.

Like many technology start-ups, ViSenze faced challenges in recruiting software engineers skilled in machine learning and computer vision, particularly in Singapore, which has a limited pool of software engineers.

“Finding engineers—especially those with R&D (research and development) experience and some specialisation in computer vision — was difficult to begin with.

“We set very high recruitment standards, which made things doubly hard.

“Fortunately, we have been able to attract well-qualified scientists and software engineers who love what we are doing and are passionate enough to make a difference with us,” Mr Tan says.

Having support from Spring Singapore’s Technology Enterprise Commercialisation Scheme (TECS) helps too, especially during ViSenze’s first year in business, when Mr Tan and his colleagues were bootstrapping and piloting its prototype with overseas trial customers like Rakuten Taiwan.

Mr Tan says: “The TECS helped us to defray a sizeable portion of costs in overseas travel, manpower, equipment and licensing. This gave us more runway and room to pilot, refine our tools and continue our core R&D activities at the same time.

“As a hard core machine-learning company, it is especially important for us to continue our R&D to sharpen our edge over our competitors.”

By Aaron Tan

Source: The Straits Times © Singapore Press Holdings Limited. Permission required for reproduction