The iconic sound of a photograph being taken.
Today, most of us don’t literally snap a picture anymore. We click a photo.
And with each image taken by people comes an opportunity for brands and organizations like yours to learn something about these people.
With market research tools and social media analytics platforms growing in their intelligence and capabilities, we can now conduct research on the more than 3 billion images uploaded to the internet each day.
Marketers, advertisers and market researchers can do the most sophisticated consumer research at the drop of hat. With tools like ours, Infegy Atlas, it’s simple to research both text and image posts.
As we become a more visual, social society, it’s time to make sure you’re up to date on all the advancements being made in image analysis.
So, who should we ask?
I sat down with Ravi Patel, developer and image analysis maestro here at Infegy, for a little Q&A to help explain it all.
First, meet Ravi. He’s a cool cat. And drinks as much coffee as I do.
Below, you’ll find in the Q&A that he’s got some fascinating insight on social intelligence platforms and image analysis. But first, let’s briefly go over image analysis, what it is, and where we are now:
What is image analysis?
Simply put, image analysis is a function of computer systems that can process images, either directly from a camera or fed to it by a hard drive, read what they see, diagram and analyze before feeding back information of that image or images to software systems.
For our purposes in marketing, market research and social media, image analysis is the method of conducting research about people by analyzing visual content on the web.
This process is accomplished by using artificial intelligence to read and process an image and recognize and identify shapes, features, and products within an image, even without text to provide any hints.
Image analysis can help brands:
- Improve market research to better understand audiences
- Build better buyer personas by improving the amount of data
- Analyze and measure their performance on social media
- Build better campaigns to more accurately reach their target audience
- Understand the use-cases of their product and the types of experiences consumers are having
Now, let’s get to the Q&A with Ravi.
So let’s start with this: how did we get here? What is the history of being able to use machines to analyze images? How new is this type of technology?
Image analysis isn’t a new thing, actually. The concept of computer vision has been used for decades and the model in its current existence has been around since the 1990s.
What’s changed is us. The way we communicate, the amount of images, the quality of images, all of that has helped our computers process imagery better. Now, it’s about us using the technology in new and different ways.
What is improving these machines capabilities? What innovations are happening to allow for them to get better?
The hardware itself plays a huge role. As machines get smarter and faster, we can program them to do different things. The evolution of image analysis has been helped by a few other things.
Camera quality, image quality, the rate of adoption, and higher use of visuals, all of those have helped improve the quality and quantity of data that these machines can process.
If you think of an image as a bunch of RGB picture values, colored pixels, and you think of a giant matrix with these color pairings input into it, the GPU can take that matrix, all of that information, and process it and turn it into algebra, essentially. The various algorithms processed by image analysis computers helps make out shapes, figures, items, products and logos.
The strong emergence of graphic processing has enabled platforms to process images better.
Why is it important for brands and researchers to be able to analyze and research image content? What advantage do they get over purely documenting text?
As we think about people visualizing their buying and consumption habits, we can use image analysis to figure out how people act and what they’re about.
The challenge then is to figure out what image analysis is telling us.
If a user posts an image that includes a can of La Croix, how do we know really that they’re actually endorsing the product or if they just actually have it and are drinking it?
We can build a large audience graph of images and that can tell us how much consumption is happening in the market. That’s a first step.
The appearance of brands within images could show us how much audiences display their products and what kind of splash they’re making in the market. But as the computers get better, we will be able to learn more.
What’s the rate of adoption for utilizing image analysis for brands and market research?
It’s becoming much more mainstream. It first was used popularly in things like detecting license plate numbers and in security systems, but now we’re feeding images from all over the globe and placing them on the internet for platforms like ours to find. Brands are catching on too, which is exciting.
What’s next? What advancements are coming?
The machines are getting better; they’re getting faster.
The raw technology themselves can be used in a lot of different areas.
We’ll see this used in scientific research and studies such as predicting Alzheimer's or better visualizing CT scans so we basically can even replace the need for having a specialist on hand at all times using certain medical devices.
Also on the horizon is gleaning more information about an analysis, such as reading emotions in an image. This could help brands not only analyze people who are posting images of their brand or logo, but also how they feel about it, even if there isn’t any text.
This technology is also spreading outside of brands and marketing.
EMS systems, for example, could use them to detect when an accident has happened at an intersection and dispatch emergency crews automatically.
Brands are thinking more scientifically too. You think about a makeup brand trying to optimize their products for certain skin types or deodorant brand trying make their products last longer. Image analysis can help with thing like that as well.
Why can brands not afford to look away with regards to this technology?
This is a new frontier that we don’t completely understand, but it would be important for every brand to keep this on their radar. You don’t want to be left behind when it comes to this technology.
What does the future hold? What are the limits of image analysis technology and will we reach them soon?
Our imagination is our only limit. That alone can make us dream up some pretty big things. We’ve only scratched the surface with image analysis technology.
Computers can see deeper and in different ways than we humans can. We could add things like infrared or other formats cameras can see in ways our eyes can’t that make the opportunities almost endless.
You look at self driving cars, which use image analysis to help drive the cars. Well, the information we get from those vehicles could help lead us to even remap our cities and make them more convenient and safer.
The tertiary effects of this technology could be anything really.
What is Infegy Atlas up to with its image search? What is future of our platform and what can users get excited about?
Our next steps are continuing to analyze the rate of adoption and use cases so we can improve the platform and have stronger integration in our systems.
We’re also experimenting with new ways to implement. If we can analyze the trends of specific logos within images in certain clusters of geography, for example, maybe we could understand better insights about people’s location using image analysis.
One example: some studies done using image analysis found that Pabst Blue Ribbon beer logo was popular in and around the Portland area. What those audience researchers found was that this was happening because of beer hipsters. Of course!
The beer brand had previously targeted older people and those living in northern or more rural areas.
By applying this learning, the CMO managed to help bring PBR back nationally, they also improved their messaging by saying “look, it’s actually cool to drink PBR, people in Portland are doing it!”. So that’s a great example of some of the capabilities brands could have leveraging this technology.
Any other things we should watch out for in the near future?
Think of all the crazy good things we could do with having computers be our eyes for us.
This could help us with law enforcement, it could help us develop smart cities, we could have things like trash collection optimized to have our trash picked up when it’s full, we could have better control over the climate of our cities if computers show us how to better build our cities to be more energy efficient.
Like I said, our imagination is our only limit.