Why Do People Unfollow Brands on Social Media? How to Keep Your Followers Engaged
By Emily SmithNov 28
Published October 3rd 2022
Brands can use image detection to find out where their consumers see them. Without this information, brands can be blind to the torrent of threats and opportunities directed at them each day. Once you have these images, you can conduct analysis, better understand your brand, know your customers, and improve your marketing strategy.
The internet speaks in imagery as well as words, so it’s not enough to just search for keywords in online conversations. Protecting brand copyright, seeing where your images are being used (and misused), and widening your e-commerce network are all good reasons why you might need image recognition software.
Without image detection technology, you won’t see the 80% of images that don't mention your brand in text. You might as well be going in blind. Even better, these tools can do that work in no time.
By using one of these platforms, as a brand manager or commercial director, you can gain deep insight into how your brand or product is being presented online – and what issues you may need to tackle.
If you are an image recognition rookie, this blog will come in handy. When looking for a tool, it’s a good idea to try a couple of them and choose the one that suits your needs best.
To start off, here’s six key questions to ask yourself before choosing an image recognition tool. They all have strengths and weaknesses, so this will help you work out exactly what you need.
Flexibility and choice is clearly important. Some services will have a limited number of logos that you can search for, while others will allow you to choose whichever one you like (including logo variations).
Logos can often be obscured or tiny in pictures. Find out if the tool you’re looking at can handle these situations without failing to detect your logo.
Some services can take a long time to detect a logo (five weeks, in some cases). Speed is important, especially for real-time conversation tracking.
Other tools can track your logo as a quickly as a few hours or a few days at most (such as our Image Insights offering).
A false-positive is when a logo is incorrectly detected in an image, such as a tool thinking a logo is present when it isn’t. Be sure to investigate this when choosing your tool as false-positive rates will vary from technology to technology.
Being able to compare and contrast data from both image and text mentions in one place is important. It means you can get the whole picture, while seeing how the two types differ.
Some platforms can only identify image mentions in relation to specific keywords provided by the user. This means you’ll miss a lot of mentions.
This is why you need to invest in a platform that gives you complete access to the mentions of your logo and those of your competitors, even when there is no text attached. The ability to logo search by image alone is nothing new, but the technology is always getting smarter, and it’s now possible to find images that only feature a small slice of a logo.
Searching images for logos is usually the first step marketeers and brand managers will take into image recognition. This is the bread and butter for understanding how a brand is being presented online. A company’s logo will pop up everywhere – from its own website and social channels, to review sites, news articles, and even competitor domains. You’ll likely even find mentions you weren’t aware of.
On top of this, you might be looking to track unexpected places your brand appears, base a campaign around visuals, or make sure your product photos are being properly picked up by search engines with product image recognition.
In the past, you had to rely on the alt text within an image to find it. This meant relying on somebody to actually add alt text when uploading the image, which doesn’t always happen. Software development and AI integration over the past decade means image recognition is now possible without any text attached. This opens a whole new world of available analysis, something every marketer will be keen to dive into.
Image recognition is becoming a competitive field, and there are tools to help you organize, analyze, and prioritize visuals in the way we currently scrutinize copy with social listening tools.
In this blog we’ll go over the top image recognition tools, and pull out what will most suit a brand’s needs.
Google has developed one of the quickest image recognition platforms available to web users. Updating your knowledge of how powerful the familiar tool can be is a great starting point for anyone seeking a quick snapshot of how far and wide their brand or products have spread online.
Google’s image recognition tool classifies still and moving images, and has been fine-tuned since launching in 2014 – a move away from reliance on alt text. It picks up not only image use on websites, but crucially on social media too.
Google Image Search is a free tool. You can access it by clicking on Google’s 'Images' and then using the camera icon in the search bar.
Google has also launched an image recognition app called Google Lens. You can read more about Google Lens below.
It's no surprise that we're mentioning our own tool here, because we believe our in-house engineering team has developed world-beating image recognition software.
Brandwatch has been at the forefront of image recognition since the team first built our software back in 2017, and it’s ideal for marketers and brand managers. You can use it to collect and analyze images containing your brand, understand your audience, and spot good (and bad) trends before they go viral. Brandwatch Image Insights is a world-leading tool that’s optimized for brands and agencies.
After independently benchmarking the technology, our clients found it to be:
Amazon dove into the image recognition game in 2016 when it launched its Rekognition tool, which primarily focuses on facial recognition when analyzing images and videos. Because of this, Rekognition has been sold to security systems and governments all over the world.
But Rekognition can also be used as a logo identifier. It uses deep neural network models to detect and label thousands of objects and scenes in your images. However, it can only analyze the images you provide, meaning if you wanted to search for the Nike logo, you would first have to feed it with thousands of Nike-related images.
If you’re using recognizable figures like celebrities or influencers in your campaigns, it’s also very handy for picking up images and videos they appear in.
Developers and researchers use Clarifai to build apps and manage data. Clarifai is one of the most accurate, out-of-the-box image recognition APIs that helps you tag, organize, and interpret data. It can sift through unstructured image, video, text, and audio, and its software lets you organize the entire data set.
This can be very helpful for market research, if you’re trying to understand how a topic or audience spreads online, or getting a feel for a new subject matter. You can also moderate content quickly and easily by teaching its AI to recognise what you don’t want to see.
You can test Clarifai's image recognition platform with a free API plug in, to see just how powerful the tool is.
Google is going to appear a few times in this article, and Google Vision AI goes a step further than simple image search. Machine learning allows you to either train your own custom image models, or use Google's pre-trained platform.
The idea is to efficiently categorize and store thousands of images, and enact quality inspection and product searches with ease. Just plug in your image sources and Vision will analyze them and tell you what they're all about.
Each image is assigned labels, similar faces, and objects are categorized together, and you can even see how visible your image will be on Safe Search. This means you can analyze your own and your competitor’s content to see what different audiences respond to, or get a feel for what a new audience likes to share.
You can try out the tool for free, just like we’ve done with a picture of a beach below.
GumGum provides brands and advertisers with a smart platform to create in-image advertising. A few years ago it developed an image recognition tool that can find photos on the internet – including social media – that are relevant to your brand.
It means a brand manager can find images on social media platforms like Twitter or Instagram that include their branding or product, and add it into their own campaign, saving a lot of time in the process.
GumGum's brand identifier software is part of its overall package of ad placements and analysis for both advertisers and publishers.
VISUA started out as LogoGrab, a company formed by ex-Google employees who realized brands were falling behind on their visual marketing and copyright protection. They created this powerful software that identifies marks and logos within images. The technology is so powerful it can even find parts of a logo and discover logo misuse.
In 2020, the company expanded its tech suite beyond brand logo search to include Visual AI. Now, alongside its logo identifier, VISUA can conduct visual search, text detection, custom object detection, and object and scene detection. This is brilliant for tracking images that might have been modified as part of a meme or with filters.
Brandwatch and VISUA have partnered recently to develop a platform perfect for social, but their own proprietary technology is world-leading for image and video discovery.
IBM’s image detection technology is found within the IBM Cloud package, and helps brands understand the contents of images, useful for analysis of previous campaigns and marketing activity. For example, it can find human faces, approximate age and gender, recognize food, act as a brand identifier and find similar images in a collection.
What's more, you can classify images and create data sets to better understand your content. Sounds great, right? Brands can also train the tech by creating bespoke detection to find a dress type in retail, identify spoiled fruit within an inventory, and more.
IBM offers its technology across three tiers: Lite, Standard, and Premium. Even the Lite version allows you to detect 1,000 images a month on custom and pre-trained models.
Imagga is an all-in-one image recognition platform that excels in categorization. Its API can be customized to generate instant image organization data that marketers can take with them when planning strategy.
Facial recognition, NSFW adult content moderation, and visual search also forms part of the Imagaa package, meaning you can cut through the noise to find images that really resonate with what you’re trying to discover.
But perhaps the most practical assets in Imagaa’s suite are the Cropping and Color features, which can transform your images based on findings from your already-sound research.
It’s not flashy, but Filestack can handle big batches of images and swiftly process, tag, and categorize files into understandable groups. It integrates with file-sharing services, so even an image recognition newbie can turn over reams of images – say from an event or new client, and have them categorized at the other end.
Quick and efficient, Filestack offers various image upload methods, while its Multipart Uploads feature allows users to upload images in smaller, more manageable chunks. If you’re dealing with vast numbers of images for brand or marketing campaigns, it will save you huge amounts of time.
Microsoft’s Azure Custom Vision is one of the most flexible image recognition software out there. It requires you to build your own model from the starting blocks of only a few images. From there you can let it run free using its machine learning capabilities, that can do things like categorize images and act as a logo finder by image.
Users can rely on the flexible Azure Custom Vision interface to navigate through this software even if you have no formal training. After which you can then relay the information to other devices to run real-time image recognition.
Microsoft also lets users deploy Project Trove – an exchange connecting AI developers to photo takers – to collect your images and feed them into Custom Vision. Form recognizing software, facial recognition, video indexing, and content moderating are also available under Microsoft’s Cognitive Services umbrella.
Detailed image labeling is vital to ensuring vast amounts of images are comprehensively categorized correctly. And V7’s AI annotation software enables users to quickly categorize images within images, spot quality issues, and save hundreds of hours of brain power.
In-person annotation of an image within an image – for instance, weaving a line around an object like an apple in a bowl of various fruits – takes on average 42 seconds. With V7, it takes 13 seconds. If you or a staff member has 10,000 images to annotate, that’s a lot of seconds saved.
Perfect for designers, the medical industry also uses V7 to annotate X-rays. What’s more, you can store all your info in V7’s Dataset.
Visio is a video recognition software that businesses can use to spot day-to-day patterns and make efficiency savings. For example, recognizing pedestrian sidewalk congestion at specific junctions, or why more cars use a particular pump at a gas station more than others.
Its powerful human recognition software monitors, updates, and produces data that leads to actionable resolutions. This could be handy for tracking your OOH campaigns, or monitoring an event you’re organizing.
Superannotate is the ‘other’ image annotation software that is often compared to V7 – and the technology is effectively the same. Efficient image, video, and text annotation means you can build high-quality datasets without the need to trawl through every item.
From there you can understand trends within your images, videos, and written text, organize into campaigns, and see which images are more successful. It’s brilliant for analysis.
Like V7, you can oversee quality management projects so the datasets you’re delivering are ironed out perfectly.
The retail industry handles billions of product images every year and even small companies can struggle with categorization, even more so since COVID-19 drove more business than ever online. Vue.ai helps retail firms develop their content management strategy by moderating image quality, product tagging and taxonomy.
Combine this with Vue.ai’s Customer Experience Management system, and its Retail Automation software – which lets brands create virtual dressing rooms and undertake product photography automation – and you can transform how you manage and expand your e-commerce site.
Syte is another retail-facing platform that uses its image recognition tool to help shoppers find the products they want to buy. Deep product tagging and descriptions swiftly boost an image’s searchability, so that it stands out over thousands of similar images.
From here, marketers can use the Visual Discovery Suite to identify retail items within an image and match them up with their products. Using camera search, the visual AI recognizes products, gender, age, and more to direct shoppers to what they’re really looking for. This is important in an ever-loudening world of online shopping, to help your customers find your products as smoothly as possible.
Google Lens sits in the Google.com app and is a shopper’s dream, offering ‘reverse’ image recognition, by providing data related to an image. You can quickly snap an image of a handbag someone’s holding on the subway, or a pair of sneakers, and see both it and similar products available to buy online.
This is customer-facing image recognition that goes much further than retail. Translate languages, take photos of text, answer ‘what is this logo’ questions instantly, learn about landmarks next to you, and even take pictures of menus to get a picture of each dish.
It’s also possible to search the saved images on your phone via Google Lens, which bypasses a phone’s basic AI that groups images together in folders. Understanding how your products appear on this app is an emerging part of SEO, and one that could soon transform how companies present images and logos – from color and detail in brand artwork, through to the models used in photoshoots.
It’s pretty clear that the image recognition space is a crowded one.
New startups are popping up almost every month with a new interesting development. Our suggestion is to try out as many tools as you can before making a large investment. The right tool is out there, you just have to find it.
If you are interested in seeing what Brandwatch Image Insights can offer, then please get in touch by booking a demo.