can we really find what we’re looking for with only one image? Yes, thanks to Deep Learning one of the latest and the most efficient branch of Artificial Intelligence (AI).
Today searching for the perfect outfit is becoming a real challenge. We, as customers, have an endless range of physical to digital shops to visit before finding our dream products.
Of course, by the time you have scrolled through the 30 pages of ASOS, checking out all the 1,034 red dresses in the results’ page, your mind is already on a new fashion trend and BOOM you’ll want to buy something completely different.
That is exactly what your customers are thinking while they’re scrolling on your website. Unlike my persistence to scan through all the e-tailers webpages, ordinary people would just click on the red cross after 30 seconds of frustrating search.
BOOM you lost a visitor, a potential client, a loyal customer.
Strangely you realize that your conversion rate is decreasing exponentially, converging towards the biggest players of the industry: Amazon, AliBaba, Asos… or simply the physical shops.
Welcome to the See-now Buy-now Era, where the client wants it A.S.A.P.
Because the customers’ behaviors change with time, so should the User Interface and the features, to enhance the whole User Experience.
We realized, thanks to the See-now Buy-Now, that people didn’t have to type a query to find what they were looking for, and when most of the time they don’t have words to describe a product seen on someone in the street, the best would be to give the possibility of providing eyes to the search.
Already in 1911, Tess Flanders was saying « Use a picture, it’s worth a thousand words ».
And indeed, everyday 90 millions of pictures are published on Instagram. Isn’t it obviously a proof that it is a new powerful communication medium?
But the real question is: can we really find what we’re looking for with only one image?
Yes, we can thanks to Deep Learning one of the latest and the most efficient branch of Artificial Intelligence (AI).
How does it work?
Deep Learning is one of the most recent branches of AI. It represents a set of algorithms which are capable of learning by themselves after being trained with relevant examples.
In order for a machine to identify something very specific within an image, it must first understand what we are looking for.
Let’s say, we are looking for a Louis Vuitton black Alma bag. First, we need to show a set of images in different contexts to the machine, enough to create its own representation of the Alma bag. This is achieved through a semi-supervised learning process.
Saying this, the top of mind use case for this state-of-the-art image recognition technology appears to be visual search, or what we call “the shazam of fashion”. Who hasn’t dreamt of snapping a look in the street and buying it online? Well after dozens of attempts, it is now possible thanks to scientific progress.
But it’s even possible to go further.
As the machine is able to detect fashion features on your product catalogue, it is also able to detect the premises of upcoming fashion trends on social media. Meaning, the machine is able to follow and pull up trends found in the content generated by influential users and push matching items from your catalogue to your consumers.
That’s exactly what Heuritech is doing. After training its machine with the best fashion experts, the startup’s technology will make your ecommerce website even more trend-driven and agile regarding fashion’s volatility.
Guest post by Marina J. Ali, Marketing Junior at Heuritech.