Translate Visual Content with Image RecognitionMay 17, 2017
An image today might be worth more than a thousand words. Already, more than 1.8 billion photos are being shared across social media every day, and researchers conservatively forecast that 1.2 trillion photos will be taken in 2017.
While Zoomph has long helped marketers to decipher text and emoji-based conversations, our newest feature lets you read visual content. too.
Powered by Amazon Rekognition and Google Cloud Vision, Zoomph’s image recognition lets you analyze the objects, expressions, and logos in your consumers’ photos to help you capture their interests and understand the full story behind what they’re saying about your brand.
How to Access Image Recognition in Zoomph
Image recognition is available to any customer with an annual Zoomph package. For more information on add-on pricing, please contact us.
Once we’ve enabled Image Recognition in your account, navigate to “Content Discovery” or “Moderation” to filter your feed based on photo features.
Zoomph’s image recognition engine will automatically tag posts based on the objects, facial features, emotions, genders, and/or not-safe-for-work (NSFW) content it identifies inside attached images. For videos in your feed, our engine will create tags based on the cover photo of each video.
You can view a full list of these tags in the right-hand filter menu of your dashboard. Under “Image Classification,” you’ll see all the tags applied to the images in your feeds. (On the contrary, you won’t see any tags that aren’t applied to one or more pictures in you feed.)
Click on any tag(s) to filter content based on image criteria.
From your “Content Discovery” or “Moderation” dashboard, explore content on a post by post basis. View the different sentiment and messages associated with images, and/or contact individual users based on what was posted.
By clicking on the image, you can get a more detailed view of the post. From here, you can also manually add or remove any tags as you see fit.
Navigate to your “Overview” page for an aggregate report of the images in your feed. You can also go to any page listed under “Insights” in your left-hand menu for more comparison reports.
Note: the demographic, ethnicity, or generation data that you see on these pages are specific to the author who posted the images, not the people captured in the images.
However, you can navigate to Content Insights > Tags to get a roll-up KPI report of the different types of images in your feed. In this report, you can see how many social posts included images with boats, water, or any other features that were detected. You can simultaneously view other important metrics, such as ZPoints score, impressions, or people engaged by these social posts.
This report is helpful for spotting trending types of images among your audience. You can gain a sense of whether images receive a positive or negative reaction, and use this as a starting point for segmenting your audience by their interests.
Cut Out NSFW Images from Your UGC Visuals
The beauty of real-time UGC marketing is that customers and fans can earn instant recognition for sharing about your brand. But when left unattended, UGC visuals can quickly turn risky—or risqué—by nature of being an open forum.
Though you can moderate exactly what gets shown on your Zoomph visuals, you may be short on time and choose to automate moderation – in which case, image recognition will come in handy.
With image recognition, you can set up workflows so that any inappropriate or offensive photos are left out from your visual. Our image recognition engine will detect lewd gestures, nudity, and more, and do the dirty work of removing them from your feed.
Track Logos Across Social Media
It’s easy to miss conversations around your company when your brand isn’t directly mentioned, or your hashtag isn’t used.
Even mega brands, like Coca-Cola—who pump several billions of dollars into advertising—know it’s a pipe dream for every person to tag #ShareaCoke when posting a picture that features their classic red can (or bottle) in hand or in the background.
But logo detection opens the door for these pictures to be found. Using logo detection, you can discover unbranded mentions of your product(s), stores, or services. You can also keep an eye out for objects or other logos associated with your brand to drive conversations around partnership opportunities, giveaways, or promotions.
Keep in mind that not all brand mentions may be positive. Some unbranded photos or messages may include negative feedback or mockery, which still renders image analysis critical to your day-to-day operations. You can put out flames before they evolve into virtual firestorms, or detect when a marketing campaign needs changing up.
Co-founder and CEO of Peppercomm, Ed Moed, also told SocialTimes about how images helped his agency track misuses of brand logos and analyze whether brand messages were reaching the right audiences.
Segment Your Audiences
Text analysis simply isn’t enough anymore. The boom of visual content requires you to analyze the images people share on social media to fully understand their expressions, interests, and overall social identities.
As an example, your consumers might brand themselves as adventurists, which may be reflected in the number of outdoorsy or active photos they share on Instagram or Facebook. Beyond that, a group of adventurists might identify as hikers, rock climbers, or fishermen.
Whether you’re targeting foodies, pet owners, or family-oriented users, tracking visually expressed interests helps to produce rich, narrower insights on different pockets of consumers.
Serve Relevant Ads
To increase its competitiveness, brands like Under Armour have already started targeting ads based on the images consumers are posting to social media.
Under this practice, if you post a picture of you training for an upcoming marathon or sporting fresh kicks, you might soon see an ad around running apparel, running gear, or a runner’s training camp.
Contrary to being creepy, these hyper-personalized ads have effected conversions and click-through rates five to 10 times better than industry standards.
Image recognition has also improved how advertisers keep up with evolving consumer behaviors and life stages, e.g., if you become a new parent, and start posting about baby bottles (rather than the “bottles” you used to talk about in your younger years), the ads you see will reflect that change.
With pure text analysis, it was easy to miss those nuances in word usage and meaning. Thankfully image recognition has helped to bolster social listening and to increase the relevancy of ads.