Companies like Craftsvilla and Voonik roll out image search feature for customers

BENGALURU: Like a dress that you see in a magazine or a newspaper advertisement? Just click a picture on your phone and your ecommerce app can throw up similar products for you to buy.

After Flipkart, two of India's fast growing ecommerce companies, Craftsvilla and Voonik, are rolling out image search, where the company's entire catalogue can be searched using images instead of textual inputs provided by the user.

"If the customer likes a colour or pattern, they'd want to be shown a similar kind of product and this feature is incredibly useful," said Manoj Gupta, cofounder & CEO at Craftsvilla.

Last week, Flipkart launched image search with which users can search Flipkart's fashion inventory with their own images. Fashion retailer YepMe had launched a similar feature earlier.


Behind this seemingly simple feature is a computing feat.

To recognise images and show similar results, these companies use computer vision, a growing discipline of computer science where massive computing power is used to give computers the ability to recognise images. Tech giants Google, Facebook and Microsoft have all invested heavily into computer vision. The use of computer vision is relatively new in Indian ecommerce. Until now, to throw up recommendations, ecommerce players relied on textual tags attached to various images. The other method is to look at a vast collection of historical data to make recommendations.

"That might work for other categories like electronics or books. But for fashion, visual search provides a clear advantage," said Sujayath Ali, CEO & co-founder of Voonik, a Chennai-based fashion retailer that personalises shopping experience for the user.

The firm offers image search to many of its users. "We decided not to look at aggregate data and decided to get more personal with the user right from the first," said Ali, a former Amazon executive. Voonik gets close to 2 million unique visitors per month.

Voonik and Craftsvilla, which gets close to six lakh visits a month through its website and mobile application, use the image recognition technology supplied by Chennai-based computer vision startup Mad Stree Den.

Mad Street Den's search and recommendations technology, based on computer vision, analyses an image for its colour, pattern and other attributes visually, and pulls up clothes similar to the one being viewed by the user.

"This also allows the company to generate tags automatically as opposed to needing someone to manually tag aspects of clothing," said Ashwini Asokan, cofounder & CEO of Mad Street Den.

The company raised $1.5 million in seed funding in January.

"We're not just a visual search company. We're an artificial intelligence company that has a horizontal computer vision platform going across recognition of objects, gestures and emotions," said Asokan. Anand Chandrasekaran, CTO of the company is a computational neuroscientist.

Experts consider computer vision to be one of the key building blocks for artificially intelligent machines. To make this work, large computer networks known as deep neural networks, are trained using a set of images. When the computer is shown images similar to the one it has already seen, it is able to perform a match.