in Blog

March 17, 2020

Computer Vision Applications in eCommerce

Author:




Edwin Lisowski

CSO & Co-Founder


Reading time:




10 minutes


Someone might argue that these two are entirely different disciplines. What computer vision has to do with the eCommerce sector? Granted, at first glance, not much. But there is a bit more going on around computer vision applications in eCommerce than you might think. And today, we are going to prove it! We will also talk about the future of eCommerce and computer vision technology.

Before we switch to the main subject, let’s shortly remind ourselves what exactly computer vision is. With this introduction, it will be much easier to grasp the potential computer vision (CV) has within the eCommerce sector.

Computer vision–basic information

CV is the subject that comes back quite regularly in our blog posts. Primarily, that’s because it’s potential and possible computer vision applications in eCommerce are growing every year. Today, computer vision is commonly used i.a. in marketing, IT, and the motor industry. As you may remember from our past articles, computer vision is a technology designed to work on a similar basis as a human eye. It’s all about recognizing what’s in the image or screen.

For instance, nowadays, the vast majority of cars have at least two cameras–one in the front and one in the back. They are scanning the vehicle’s surroundings in order to assist the driver, for instance, while they drive in poor weather conditions or park their vehicle. The camera monitors the view that’s in front of a car and sends images directly to the computer vision algorithm.

And it is this algorithm’s role to interpret the pictures it receives. For example, the algorithm can interpret that there’s a pedestrian at the crosswalk. If it seems that the driver can’t see them (does not slow down), the algorithm can send them a direct message (visual or sonic) to pay attention.

It’s one of the most popular and recognizable computer vision case study in eCommerce, but of course not the only one. They can similarly scan what’s happening on each customer’s computer screen, or in front of a smartphone’s camera. Probably you already know where is this going. Yes, it’s a perfect solution for online shopping, and let us explain to you how and why.

computer vision algorithm

Read more about Computer Vision Solutions.

The computer vision applications in eCommerce?

Computer vision applications in eCommerce? Yes, and in fact, it happens already! Nowadays, we can see lots of computer vision applications in eCommerce. That’s because computer vision technology can be easily adapted to the needs of eCommerce. And it works either for the retailer and their customers! Long story short–the computer vision applications in eCommerce can help in:

  • Organizing the inventory and monitoring the way it’s presented on the website
  • Finding the product a customer wants
  • Improving customer service and experience
  • Understanding what each customer is looking for

Actually, we can divide possible computer vision application into these two categories:

  1. The computer vision technology that assists the customer
  2. The computer vision technology that aids the sellerLet’s talk about the first one first.

Search by image

Currently, it’s one of the leading trends. Not that long ago, Google has introduced this new option in its Google Images search engine. Now, you can upload a picture into Google Images and search for similar ones on the web. This is exactly how computer vision works. The Google algorithms scan the uploaded picture, interpret it, and search for something similar.

In the same way, shortly, you will be able to find your favorite items by taking a picture and sending them directly to the online store’s database. And it goes further! Consider Google Lens. If you have an Android smartphone, you can check this option instantly. Just open your Google Assistant and ask it to open Google Lens for you.

What you will see is an app that uses computer vision techniques along with machine learning technologies to provide you with the relevant information about the object you’re pointing at. Google Lens detects an object that you’re aiming at with your camera, interprets it, and gives you the results. And then, you can instantly search for it in the Google search engine. This, in turn, allows you to check its model name and price.

Have you recently seen some nice shoes and you didn’t know where you can find them? Nothing simpler, Google Lens will most likely know its brand name and model. With this data, you can simply make a purchase online.

Currently, this feature is primarily used in Google, but the eCommerce companies already try to implement this solution into their businesses. For instance, in 2017, eBay announced Image Search and Find It On eBay, new possibilities for shoppers to use pictures instead of words to search eBay’s catalog[1].

Search by image, laptop

Computer Vision Applications in eCommerce–your new salesperson?

Here are other interesting computer vision applications in eCommerce. Possibly, within a few years, the CV application will take the place of a well-informed salesperson. How so? Imagine the possibility of asking this virtual salesperson questions like, “What goes well with these pants?” or “What is the style of this jacket?”

The truth is, sometimes, we like something we own, but cannot describe its style. In such a situation, it may be difficult to find something that matches our item. But no worries, computer vision comes to the rescue! Imagine that the AI algorithm can identify the object you’re showing it, determine its style, and find related products in the store’s database. Wouldn’t that be immensely helpful?

And what about communication? Currently, people don’t usually talk to chatbots (mostly because they fear it will be a waste of time). But bear in mind that Natural Language Processing (NLP) develops rapidly. Within a few years, it might be challenging to distinguish whether you’re talking to a robot or to a human being.

So, to conclude, you might end up with a virtual robot that advises you if your shoes match your jacket, or if your lipstick goes well with your skin color. Amazing, isn’t it? Although this technology is still in its infancy, we can already see some of its computer vision applications in eCommerce. One of them is strictly related to Augmented Reality (AR).

Augmented Reality

On many occasions, we see something that we like, but we are not sure if it fits our interior, our car, our wardrobe, our desk, or even ourselves. This is an especially common occurrence in the interior design niche. Thankfully, Augmented Reality is the answer. It allows you to combine aspects of the real world (your apartment, your car, your wardrobe, your desk), with computer-generated content, such as a product you intent to purchase.

The AR apps make it accessible for you, as they place computer-generated images into a real image of your room. All you have to do is to download the dedicated app to your device and choose the item you like. Next, you will see it, so to say, placed on your desk or in your kitchen. So, even before you buy it, you can decide if it goes well with your interior.

Consider another example. The MAC Cosmetics company used the new AR YouTube ads to present the latest lipstick model. Google claims that during the test campaign, about 30% of people who displayed the ad, appeared to launch the AR platform, and then spent on average over a minute with a “virtual lipstick”[2]. If you want to test this technology, open this link on your smartphone.

Just below the recording, you can see the “try it” button. After clicking, the front camera of your smartphone starts. You have to point it at your face, and then, you can check out the different shades of lipstick on your own lips!

Augmented Reality, boy

How can computer vision technology help sellers?

Now, let’s talk about sellers for a while. As it turns out, CV can aid them significantly as well! First of all, computer vision can be used to maintain high-quality descriptions in your store. How come?

The computer vision applications in eCommerce can regularly scan your virtual warehouse and correct any product attribution mistakes, or add missing visual product attributes, such as color. A deep learning algorithm can flag the potential error and mark it as a computer-generated notification. In turn, human employees can verify the mistake and, if needed, update the given product’s description.

Another interesting computer vision applications in eCommerce refer to the delivery bots. The company called Segway introduced delivery robots at CES 2019 in Las Vegas, USA[3]. According to the company, these delivery robots are fully autonomous and can even navigate through obstacles and crowded environments! They can take packages and letters from a mail carrier, then walk on their own to the right offices or rooms. The company says that these Loomo Robots are equipped with interactive screens, a cloud-based service dispatch, real-time monitoring, and elevator functions.

And last but not least–we’ve already talked about virtual warehouses; now it’s time to tackle the real ones. In the not too distant past, human employees were working really hard to ensure timely delivery. They were also responsible for moving and placing products in the warehouse. Now, more and more often, robots with computer vision algorithms do the majority of these tasks. The robotic arms can detect, classify, and read the exact location of a particular package, grab it precisely with an appropriate amount of force, and move it to the designated location. It’s a faster, cheaper, and more efficient solution.

seller, woman, man, store

How will computer vision applications in eCommerce change the sector in the future?

What about the future of computer vision? Actually, it’s right here. We can expect all of the aforementioned solutions and applications to multiply within the coming years. Surely, in the eCommerce future, we will see significant improvements in the virtual customer service, particularly with the usage of chatbots, which will be more human-like. The AR technology will be more and more advanced, and 3D computer-generated items more realistic. The search by image option will be more advanced, and its recommendations more accurate.

You can also draw another conclusion from our deliberations. Ecommerce automation is and will be a growing element of online business. This means that human employees will have more time for other duties. For example, companies that possess large warehouses will invest more and more money in robotics to make the warehouse functioning more efficiently.

And finally, there will be much more importance placed on mobile devices that buyers use when they’re shopping online. That’s because they are simply faster and more convenient than traditional desktop computers. We talked about Google Lens–this function is available solely for android devices, and most likely, other search-by-image algorithms will go the same, mobile path.

As you can see, the eCommerce future is bright and dynamic. Indeed, you don’t want to be left behind! Be in no doubt, we can prevent that from happening! We encourage you to find out more about computer vision. Are you keen to implement this technology into your online store? No wonder, it’s a real game-changer! Give us a call and let’s talk about your needs and ideas. The Addepto team is always here for you!

References

[1] Ebayinc.com. eBay Powers Searching and Shopping with Images on Mobile Devices. Jun 26, 2017. URL: https://www.ebayinc.com/stories/news/ebay-powers-searching-and-shopping-with-images-on-mobile-devices/. Accessed Mar 17, 2020.
[2] Jon Porter. YouTube’s AR ads let you try on virtual makeup alongside beauty vloggers. Jun 19, 2019. URL: https://www.theverge.com/2019/6/19/18691102/youtube-augmented-reality-ar-ads-beauty-vloggers-mac-cosmetics-lipstick. Accessed Mar 17, 2020.
[3] Marrian Zhou. Segway to introduce autonomous delivery robots at CES 2019. Jan 3, 2019. URL: https://www.cnet.com/news/segway-to-introduce-autonomous-delivery-robots-at-ces-2019/. Accessed Mar 17, 2020.



Category:


Computer Vision