What is image analysis?
Something we can't live without. Images grab attention. They’re memorable. They’re engaging. They’re shared. Big time! Should brands care? Yes. With over 3 billion images shared online every day, how can brands police their logo? Protect their reputation? Boost awareness? I’d like to talk to you about image analysis. The technology as it was. As it is. As it will be.
Here it is...
80% of posts where brands are detected in images, make no text mention of the brand.
You're missing all that social data about your brand!
What is image analysis?
Here's the definition of image analysis. It's the process of taking information from images. It can be as simple as scanning a barcode, or as complex as PiP.
Yep… one of the most advanced pet identification systems out there...
PiP is a smartphone app created for pet owners who’ve lost their cat, dog, fish. Register a photo of your pet in the PiP app. Should you misplace your pet, its photo will be analyzed and matched with photos of pets that have been found wandering the streets.
Image analysis is used to beat lost tags, outdated microchips, and fading tattoos.
Teaching a computer to see, is no walk in the park. For your machine to see as we do, it uses computer vision and image recognition. Computer vision is the processing part, not the eyes. It interprets what a machine sees.
AI-powered image recognition uses computer vision to make sense of input received so it can categorize. Real-world examples include…
- Driverless cars - use a combination of image recognition and computer vision to identify other vehicles, pedestrians, and road signs.
- Price comparison - consumers photograph something they’re interested in buying, and use an app like Google Shopper to find nearby stores and prices.
- Google Image Search - present an image or URL and the search engine will show you all instances on the web where the image has been used.
- Is it a dog or not? - Not Hotdog - for those times when you really need to identify a… hotdog.
It’s a bird! It’s a plane! It’s a hotdog!
Training a neural network - the part that processes pixels in an image - is complex. But, show it a pre-labelled image enough times, and it will learn to recognize similar images.
What’s the difference between image analysis and image recognition?
Image analysis definition - it's for an image you’ve already stored. It identifies what’s there. For example, the image below would be tagged with one man, two pairs of spectacles, three laptops, table...
Image analysis identifies what can be seen in the image.
Image recognition definition - it’s magic! It’s AI-powered tech that finds images online that contain a pre-defined search query. In the example below, the query is Talkwalker.
Hey presto! Two instances of the logo found in the shared image.
Should image analysis be part of your marketing strategy?
It’s a vital part of any marketing strategy, due to the huge number of images posted online every day. Over one billion containing brands and products. Talkwalker’s Analytics platform offers image recognition and video analytics. Identifying multiple elements, then comparing to our database of 30,000+ logos, scenes and objects.
Does this mean text analysis is dead?
No. Not at all. Never.
If you want the whole picture - everything that’s being posted about your brand online - you have to use text and image analytics.
With visual content being shared like crazy, it’s a no-brainer. Visual listening is the way forward for brands.
How will image analysis help my brand?
First up, let’s look at where you should search for your logo...
- Selfies - what’s the sentiment - smiley face, grumpy face?
- A photo with an influencer - a celebrity would be cool
- A photo in a particular location - beach, bar, kitchen, bedroom
- A photo of a billboard, real-time digital sign, sponsorship banner, street furniture - bus shelters, phone booths - transport ads - taxis, buses, subways, trains, guerilla advertising
There are a heap of things you can do, if you employ social media analytics, specifically image analysis.
Track brand mentions accurately
In the past, it was all about mentions of your brand and products in social media posts. Text-based and/or direct tags of your brand. But, this method means you miss so much. What about the posts that don’t have any text?
Just an image. Image analysis will find all instances of image-based user-generated content (UGC). Including visual listening in social media analysis, means you can identify images that include your logo and product. The positive ones, providing UGC to feed your marketing strategy.
Talkwalker image recognition identifies brands, mentioned in content or not.
no es la gloria esto?? pic.twitter.com/DQYLMZ6WJ3— әu!ɯzɐɾ (@_jazminalmiron) February 27, 2019
Yep, it’s glorious. Glorious that image recognition found this positive post that includes Coca-Cola, but no mention.
Smarter online merchandising
Visual product discovery. Consumers choose a product they’re considering buying. They’ll then be shown similar products. How is this beneficial? For an ecommerce site with a large and varied product list - Amazon - it means that more products can be highlighted. If you search for trainers in the fashion department, product discovery would also offer trainers from the sports department.
Brand reputation protection
Again, we’re talking about shared posts that picture your product, but don’t reference in the content. Sentiment analysis will read and understand the words, and tell you how consumers feel.
Image analysis will identify your brand or product in the image.
Negative content going viral is not good. Negative visual content is even worse, as content with visuals is 40x more likely to be shared than other types of content. Crisis management is crucial if you want to identify potential crisis triggers and protect your brand reputation. Find them with visual listening.
Prove the value of sponsorships
How do you know if paying a heap of money to place your logo in a sports stadium, is worth it? What about that huge event you just sponsored? What did it bring to the business? These are questions that your boss will assume you can answer… immediately.
What was the ROI?
View this post on Instagram
Jody Scheckter in his Ferrari 312 T4 at Monaco (1979) #f1 #formula1 #formula #1 #mercedes #redbull #ferrari #forceindia #williams #mclaren #honda #tororosso #renault #haas #manor #sauber #great #racing #awesome #moments #love #crash #start #gas #brake #clutch #shift ------------------------------------------ Follow other great F1/Motorsport pages : @gp_epicmoments @rcx.ph @f1.images @zakir_munawer @lastlapinsider
Talkwalker's proprietary image recognition technology will detect the logos of Martini, Texaco, Michelin, despite no mentions.
Image recognition technology - identifying your logo - means you can quantify the exposure your brand received. Proving the ROI of your sponsorship.
Protecting your brand from trademark abuse
Over $60 billion in profits is lost annually in the EU, due to counterfeit apparel.
Protect your trademark against abuse. Image analysis should be used to detect unauthorized use of your logo and counterfeit brands. Ecommerce sites will be able to find fake brands being sold on their websites, and brands can use the same tech to identify copies of their products being sold elsewhere.
Protect your trademark and brand reputation against damning misuse.
Find and monitor influencers
Hey look! That A-list celebrity has been photographed with your brand logo. Shame they didn’t mention your brand name.
Doesn't matter... I'm using image recognition, so I found it. Free advertising. Can't lose.
Unplanned influencer-generated content is a goldmine. If you can find it.
Discover strategic business insights
Every brand wants to know as much as possible about how, when, where consumers are using their products. Your beer frequently appears in images online. Consumers taking selfies, branded glass in their hand. What’s that? A high proportion of the shared photos show people partaking of a particular brand of snack along with the beer. Partnership opportunity?
Check out In Pizza We Crust. It’s a case study about a restaurant chain that used image recognition to determine where in Europe to expand the business. It revealed where people like to eat - inside or al fresco. What they drink with their pizza - wine, water, juice, beer. What toppings they enjoy - you get my drift. It’s a slice of genius.
Capitilize on user-generated content
Exploit user-generated content and boost brand awareness. It's free advertising. Don't waste it. Salesforce state that UGC can improve campaign engagement by 50%, increase email click-through by 73%, and conversions by 10%.
You can't ignore stats like that?!!
Emotion recognition technology
Nope. Not the awesome sentiment analysis in our platform. This is more to do with capturing people’s emotions on camera.
AI-powered facial analytics tech uses a webcam to capture people’s reactions when shown certain adverts and content. Faster than market research, it collects real-life, spontaneous facial reactions.
What could be better than seeing the immediate reaction to one of your blog posts, tweets, adverts, or even your product?!!
What about offline?
Image analytics is not restricted to the analysis of online content. Use cases can be found in other areas, including medical imaging - cancer detection in a mammography scan, and x-rays. Intruder detection, determining the mineral content of rock samples, defense, astronomy, etc.
No checkout required!
In 2016, Amazon Go hit the headlines. It offered, “the world’s most advanced shopping technology. No lines, no checkout - just grab and go!”
The tech involved is similar to that employed by driverless cars. Computer vision, sensor fusion, and deep learning.
Consumers enter a store via a turnstile. The barcode on their Amazon Go app is scanned, and the consumer is tracked while shopping.
Shopping completed, the consumer leaves the store. The app having recorded and stored in a virtual cart, what you’ve taken. Shortly after leaving, the consumer will receive a receipt and the cost of the purchases will be taken from their Amazon account.
Watching and counting traffic in stores is nothing new, but it's upgraded. Using computer vision means tracking consumers has advanced so much that it can be used to optimize the entire shopping experience.
RetailNext - the most advanced in-store image analytics solution available.
RetailNext uses sensor tech and computer vision so store owners can identify:
- Entrance path analytics
- Visit duration
- Unique traffic
- Pass-by traffic
- Visit frequency
- Identify traffic cycles across the day, week, or year
- Which fixtures are working and which are failing
Optical character recognition
Optical character recognition (OCR) has been around for decades. It’s a computer that recognizes printed or handwritten text, and translates into machine-encoded text such as ASCII.
In 2004, the Royal Mail in the UK paid megabucks - about $200M - for technology to scan the front and back of envelopes. It then translates the addresses into machine-readable codes, enabling next day delivery at a huge scale.
Image analysis technology is developing at a rapid pace. Who’d have thought that you’d be able to find your brand online without a text mention?!!
Image recognition is the future, happening now.
Hopefully, it won’t be long before we can perform the same task with video. Hunting down your logo in videos. Now that would be awesome!
UPDATE - WE CAN!!!!! Check out our video analytics feature. It'll blow your mind.
Looking to see our image recognition in action? Grab yourself a free demo with one of our experts.