Computer vision is here to revolutionize the way we interact with surrounding digital technology. In a few years, we can expect the mouse and keyboard to become obsolete and replaced by voice command and cameras. Experts from computer vision company InData Labs say this is a more natural way of interacting for humans, as it relies on an extension of our physical senses.
What if we could show the search engine the pair of jeans or the suit we want instead of striving to find the right words to describe it. What if we could feed a whole Pinterest board to a search engine and get back a list of shops near us with the pictured items in stock? Sounds crazy, doesn’t it? This may not happen tomorrow, but it will soon enough. Here are a few ways computer vision could enhance marketer’s efforts.
Perform Image Search
As described in the previous example, with the help of computer vision, users will be able to search by uploading an image. If the algorithm is sensitive enough, this removes the necessity to manually tag products with dozens of descriptive attributes.
This new approach will remove some critical barriers. The first is the language obstacle. Web shops which implement this tool will no longer need detailed pages for each language. It will also make shopping more accessible for those who are not very knowledgeable in specific jargon.
An essential part of marketing is looking at trends and predicting what will be in high demand. The velocity of changing tendencies is making this endeavor more difficult than ever. Marketers can’t rely on yearly or seasonal collections; they need to have almost real-time responses.
By feeding the image search data into machine learning algorithms, they can unravel essential data such as the most sought-after colors, designs or individual items. These signals can even show which products are searched for together and also purchased together. It’s a step forward from what Amazon and other recommendation engines are already doing.
Serve Relevant Graphical Content
Take a moment to think about what Instagram is doing with your home feed. They are showing you images that are relevant to your interests, previous searches and followed accounts. Right now, they are relying on text, via hashtags. Soon, we can expect the same behavior, just triggered by the user’s image search through the uploaded content.
For example, if you had been searching for a particular type of frying pan, the computer vision algorithms will generate images which are related to your initial query. For example, you might see more advertisements featuring people in cooking-related contexts. Conversely, if you had searched for a sports bag, you could expect more fitness-themed graphics to pop up.
Create Dynamic Advertising Opportunities
The situation described above is just one of the potential application of dynamic advertising. There are numerous other possible examples, both online and offline.
Online, the idea of A/B testing can be extended to a very personal level. Depending on the previous searches, not only can the person receive a customized ad version, but this can go in fine details such as using their favorite colors or depicting characters to which the user can relate.
Offline, this concept could translate into measuring engagement with outdoor advertising by tracking the reaction of people. Simple footage from surveillance cameras or specially installed cameras in the ad panels could reveal if passersby were impressed, moved or bored by the ad. With such data, the marketing teams could react much faster before an uninspired campaign harms the product.
Also, computer vision could look at the visitors from a store, extract some characteristics, like the colors or styles they are wearing and run real-time advertising on the in-store screens to match the current visitors’ interests. This idea could even be taken a step further through AR and smart mirrors.
Create Excellent Store Layout
Speaking of CCTV, there are whole terabytes of unused footage from stores which could be repurposed. By feeding it into algorithms which analyze the number of people in a specific area, when a crowd is forming it means something worth seeing is there. Store managers can identify the most exciting items and bring them forward. Also, if the computer sees people reaching out more often for a specific thing and putting it back, that could signal a missing stock problem.
Putting all this information together can create a layout as dictated by clients. Combining this with other smart devices such as RFID tags could even create a cross-selling scheme. For example, if you have already put an individual item in your basket, the system could signal through an app if a connected article is on sale. Maybe you need a new bag with those fabulous new shoes.
Perform Sentiment Analysis Onsite
Even if it seems intrusive, future algorithms will be able to scan our faces while we’re shopping or watching a commercial. This is just a step forward from what brands are already doing when they are scrape the web and our social profiles for references to their brand. Sentiment analysis is currently a sub-domain of text analysis, but this will slowly become a sub-domain of image analysis.
Just imagine what you could achieve as a marketer if you could simply ask the participants of a study to watch a clip and collect their reactions instantaneously. This would replace the boring surveys with real data that is unfiltered and unbiased.
The search world as we know it is about to change due to computer vision. Text will be replaced by images with a substantial impact on the marketer’s work. The new tools will find their use in the way clients search for items, testing advertising content and creating attractive store layouts. This will lead to increased customization of advertising and an increase in real-time marketing. Analytics and data survey tools will also get a boost from the new technology.