Major technology companies and e-commerce platforms – such as Google, Facebook, Amazon, Pinterest and Alibaba – have been investing in image recognition technology in order to stand out in their category. What does it mean for brands in the future?
Image recognition is not a new concept but it is representing a bigger share of major tech companies’ businesses.
This month, Pinterest sharpened its visual search functionality that is powered by machine learning and image recognition techniques.
Instagram is reportedly experimenting with a new ad format that users Apple’s 3D Touch feature to let users switch among multiple product options in the images. And Facebook has conducted extensive machine vision research to improve facial recognition.
As big industry players are developing image recognition to serve different purposes, how will this technology impact digital marketers?
Image recognition presents opportunities for online retailers
It’s universally acknowledged that images are more intuitive than written words. When one struggles to describe something in words, image recognition can make everything easy.
“Image recognition is so important because it has the potential to be the paradigm shift away from the written word. We shifted from images on the walls of a cave to a written word because it was more efficient. Can we make images efficient again?” says Sastry Rachakonda, chief executive (CEO) of digital marketing agency iQuanti.
The answer is “Yes,” as many e-commerce platforms are using image recognition to provide convenience. Brands can attach product details to a visual query, while users can have shortcuts for product information without necessarily knowing where a product comes from.
For example, Amazon has developed a mobile app called Firefly for its Fire Phone. The app can identify books, videos, music, email addresses, barcodes, QR codes, web addresses, and phone numbers.
If an item can be identified, Firefly creates a digital representation of the item and tracks it.
Alibaba, China’s largest e-commerce platform, unveiled a feature in May of this year to let mobile users pay for a product with a selfie.
Aside from e-commerce platforms, clothing store Neiman Marcus expanded its “Snap. Find. Shop.” function to all of its luxury departments this past September, in order to make search easier.
“Right now, it is all about showrooming and helping the e-tailers. I think the future is already here in terms of comparison shopping. The future opportunity for retailers may be in being able to harness image search to make a shopping experience a lot more fun,” Rachakonda notes.
Mass-scale image recognition could have a huge impact on search
Image recognition can be a powerful tool for e-commerce professionals because it provides an additional contextual layer for keyword search.
Pinterest has already allowed users to search for all things they don’t have the words to describe.
And Google gains a deeper understanding of search terms and a deeper recognition of content by harnessing image recognition and machine learning.
The company’s existing mobile app Goggle allows users to point their mobile at a picture, barcode or QR code and then Google will surface everything it knows about the object.
Lately, Google further created new software that can describe images in complete sentences.
Could visual search replace the keyboard on mobile?
“It’s quite possible for visual search to replace the keyboard. But text-based search and visual search coexist in different use cases,” says Omaid Hiwaizi, president of global marketing at Blippar.
For example, if one sees a dog and wants to know the breed of the dog, he or she can take a picture and send a visual query to get more information. But if one is travelling in Tokyo and wants to know the local time, he or she can hardly use visual search.
Image recognition is still in its infancy but it can be very advanced. As Google mentioned in a blog post, a deeper understanding of image can be directly transferable to the company’s products, including photo search, image search, YouTube, self-driving cars, and “any place where it is useful to understand what is in an image as well as where things are.”
iQuanti’s Rachakonda has a good point: if vision is a combination of eyes that see visual content and brain that processes it and provides context, we already have the ability to capture images better and better, as demonstrated in virtual reality, as well as process images more thoroughly. (Remember, computer Deep Blue beat chess champion Kasparove a few years back.)
“It’s now a matter of putting the technology together in a workable economic way – but visual search will be amazingly advanced,” Rachakonda says.
Image recognition enables consumers to search for objects that they don’t have the words to describe, allowing the integration of the offline and online worlds. Tech companies will continue developing image recognition and machine learning. At some point in the future, people could search without a keyboard.
How advanced can image recognition be? Sky is the limit.
by Yuyu Chen