The story behind the attractive images of Flipkart’s catalogue

Have you ever wondered the logic behind the clear attractive images that ecommerce platforms use to target the audience? Consumers shop with their eyes. A survey done by Singapore startup ViSenze found that quality of a product image is very important in selecting and buying a product to 67% of the consumers. But, it goes beyond just to attract the audience.

A leading ecommerce platform in India, Flipkart uses bright, colourful images to personalise a customer experience. No more going through hundreds of images to find exactly what you want. Whatever the occasion is, wedding, convocation, birthday, date or just a sleepover at a friend’s place, wear the perfect dress in just one click. Just upload a picture of the product you want and the product will be retrieved from the database of Flipkart, thanks to image recognition technology of ViSenze.

“A picture is worth a thousand words. But will you use a thousand words to describe the image in a search engine! Obviously, not. So why not use an image itself as a query,” said Oliver Tan, cofounder of ViSenze.

Infact, ViSenze survey showed that 74% found traditional text based queries are inefficient in helping them to find the right items online.

So, how does it all work? Let’s have a look at the whole technology.

We all love ogling at the red carpet dresses of celebrities. ViSenze used this psychological factor to target the audience. They captured images at a fashion show and with the help of computer vision and machine learning they pulled out the item which the customer would be most interested in. This image was used as a standard image to teach the machine about the attributes of the product, like if it is sari, kurta or kurti, colour, linings, sleeveline, etc. Making that image as a standard search image, they pulled out its exact match from the database.

One must not forget the noise involved in an image taken at a fashion show, like background colours, lights, and people walking behind the target item. These were removed to make it a clear image to maximise the search results. Now, if you observe the images on the site, they are clear images too without noise. What is more interesting is the clear cuts and absolute colour pixels of these images. This is done for efficient search results. And to the advantage of the customers, the whole process also makes the image more attractive.

The entire exercise was as much a challenge for ViSenze as it sounds. More than the technology, learning about Indian ladies fashion was a greater challenge, laughs Oliver.

“Talking about ladies fashion is one thing and talking about Indian ladies wear is a whole other level, with so much variety,” he jokes.

But working in the Indian market also made him realise the quick adoption rate of innovation in the country. He added that India was their top market in 2015 though they are in the US, UK and San Francisco.

Bet, after this, images won’t be just that the next time you visit Flipkart.

(Image credits: Flipkart)