In a few words, what is ViSenze ?
ViSenze is the technology that simplifies the way people look for information online or for things to buy, not by keywords but by images!
If someone wants a similar dress to one seen in a fashion magazine or on the street, then it should be easy in this era of abundance in ecommerce and smartphones with internet and powerful cameras. ViSenze is bringing visual search and image recognition to life, letting people wave goodbye to the hassle of keywords search.
After all, if “a picture is worth a thousand words”, wouldn’t it be great if we could just use the images themselves as search queries and save the “thousand words” descriptions? That is what we bring to the world.
What made you decide to build it ?
We heard discussions between our wives during casual meetings you know, the typical “Wow, I love your shoes. Where did you buy them?”. Fun aside, we did notice this big search problem shoppers have in finding what they want. Putting in words the right color, pattern or style of fashion or furniture items is not an easy thing, if accomplishable at all.
This gap between search intentions and search results leads shoppers to frustration and wasted time. And because B2C ecommerce sales are booming, growing from $1 trillion in 2012 to double in 2016 according to research, that seems like a big problem.
Motivated to solve this problem and lucky to know a few great engineering minds passionate about computer vision and deep learning, we decided to pioneer visual search and image recognition and give our APIs to all the businesses that need it, ranging from ecommerce companies to patent libraries. And today we are happy to have well known companies Zalora, Clozette, Rakuten, Patsnap and more amongst our customers.
Based on your result, how does visual search impact ecommerce?
Our customers in the ecommerce space have seen a 30% uplift in conversion rates, as well as increase in engagement, like more product views through the Find Similar functionality we provide.
For the moment, it also offers ecommerce companies a competitive edge when being innovative and powering their users with the upload-photo-to-search functionality. However, this seems to turn soon into a commodity rather than innovation. The industry already calls it “mobile search”, so you can see how common it is about to become.
Who, apart from ecommerce, would find benefits in using visual search ?
Any business that has a large image database and a search problem. For example, besides ecommerce, we are also working with a patent database, Patsnap. They needed us to improve their user experience and making it easier to find similar patents by searching using schematic drawings.
And there are endless opportunities. Publishers, advertisers, designers, brands they can all benefit from this visual technology. We are open to explore new use cases on demand.
Does ViSenze use content matching or content recognition? In other words, can it also supply corresponding plain written or spoken text to a visual query or just an image?
Our visual engine processes pixels in images and video frames to generate either similar images or descriptions of the images as outputs.
So yes, you can use an image to find out more information about it. For examples, uploading a photo of an animal can return as results information like its species, average life span etc.
Visenze offers contextual advertising . Can you explain how it works ?
ViSenze classifies and indexes visual images according to IAB standards and identifies those that resonate best with brands/advertisers. So the most relevant ads can be placed automatically on graphic or video content by ad serving engines. And all this is easily translated into better ad performance.
How is ViSenze different/better than the other visual search engines on the market today ?
Everyone will claim that they are better in one way or another. We’re not interested in comparative debates. What we are interested in is making sure our visual search technology really solves the problems of our customers, so we work closely with each of them to continuously outperform ourselves for their specific vertical. And by customers we mean ecommerce players, retailers, image libraries etc.
Our guiding product philosophy is reflected in 3 ways:
- We learn and understand the specificities of each domain we choose to specialize in (e.g. in fashion, we understand fashion apparel taxonomies and differences of visual attributes that matter to the shopper).
2. We adapt quickly from one use case to the next by using highly adaptive machine learning algorithms and redefining accuracy parameters with customers.
3. We always improve on scale and speed, currently offering processing capacity for hundreds of millions of images and results in milliseconds.
Does ViSenze has a learning capability, getting smarter with each use?
Yes, we use deep learning, a branch of machine learning and big data. Therefore, the more images we feed to the system, the better it becomes.
What are the objects hardest/easiest to recognize ?
Across the industry, facial recognition ranks amongst the hardest because faces do change and exhibit different expressions etc. The easiest to recognize are brand logos because logos are always consistent in shape, looks and color.
Will you ever be able to tell the difference between a pair of Levi’s or Wrangler Jeans ?
A pair of jeans will always be a pair of jeans when you see one. It is possible to tell the brand of the Jean if the distinctive logos or brand marks are visible enough for the algorithm to pick it up. Other than that, we can use the way the jeans look to find a matching or similar pair, with all its information, like the brand name.
What could ViSenze offer if there was no limits on technology ?
As a visionary company, we dream of possibilities of accurately offering or predicting consumer choices with high probabilities based on both visual and nonvisual behavioral signals or patterns. As a company whose technology is built on science, there are always limits in current state of science or technology. The challenge for us is to always push the boundaries harder and deeper in areas that we have prioritized.
Author: Paul Melcher
Paul Melcher is a highly influential and visionary leader in visual tech, with 20+ years of experience in licensing, tech innovation, and entrepreneurship. He is the Managing Director of MelcherSystem and has held executive roles at Corbis, Stipple, and more. Melcher received a Digital Media Licensing Association Award and is a board member of Plus Coalition, Clippn, and Anthology, and has been named among the “100 most influential individuals in American photography”