Computer vision is kicking it up a notch.
Technology has always been a booster for marketing, and the latest craze is computer vision (CV). Through this tool, machines are endowed with almost human-like senses. Not only do they “see” pictures and objects, but they can classify them, find similar items, or create an augmented reality.
These new capabilities can be exploited by marketers in different ways, including new search methods, redefining retargeting, and new methods for analytics. The real advantage of using computer vision comes from the fact that most people are visual creatures. Since the human brain is hard-wired for images and not text we can expect that these advancements will not only be embraced by users but will become the first choice.
New Search Methods
Until now the only way to find items in an online shop was to use the search bar, the embedded filters, or a combination thereof. Up to now, all searches have been based on text analysis and tags.
A new era is about to dawn, a moment when clients will use pictures instead of words to search. This trend is already visible on some websites which ask visitors to select a “preferred style” from a handful of options. Once a choice is selected, the search is further refined by smaller details, colors, or available brands. This is similar to previous category filters from online shops, but the visual examples used here are a much more powerful selection tool.
Computer vision experts from InData Labs explain that this eliminates the arbitrary tagging which could reject some items classified artificially under a different category and which would be unnecessarily excluded too soon. For example, some clothes could be both sporty and casual, and selecting just one category could leave out several viable selections.
Until now, most product tags were manually placed by a person thinking about the characteristics of the product using different elements. These included name, use, color, similar or related items, and included words or phrases the user would most likely search for in a text-based query. Unless a clear system was in place, most tagging was at the discretion of the person performing the job, as there is no recognized approach to systematic and efficient use.
With the help of computer vision, an algorithm will be able to identify items in a picture and attach the relevant categories for the shop into which it is placed. Such automatic classification relies upon fuzzy clusters. Basically, the same image can receive more appropriate tags if the objects depicted are convenient for more categories.
What if you could search for a dress or bag similar to those worn at the Oscars by your favorite star? Computer vision not only makes that possible, it will also rank results based on the best match so you can get as close as possible to your desired look.
If you’ve always hoped for a visual Shazam, you’ll love Lens from Pinterest. Using the visual search tool, you can “Lens your look” by taking a photo of an item you already own and finding looks which complement it.
Once a potential client has spent enough time looking at a specific item or range of things, this can be used as valuable information for retargeting. If they have added an article to their cart but still haven’t purchased anything, in addition to showing them that item, the algorithm can find similar objects and include these in the retargeting strategy. This enhances the chances of selling by showing more options which are relevant to the search. It’s important, however, that the items chosen for retargeting are not too divergent from the original search as this could hurt the intent.
Take Analytics to Another Dimension
An essential part of a marketer’s job is to perform in-depth analysis, identify trends, and predict the future. These goals should be part of a strategy encompassing both online and offline dimensions.
The way people behave on location can give valuable hints about their preferences. Observing them can help marketers with better product placement and store layout—ultimately impacting the bottom line. Computer vision allows the tracking of people’s movements, estimating the waiting time for fitting rooms and at the cash register, and evaluating service during rush hours. All these can be retrieved from CCTV, which is already installed, and could result in happier customers, staff that is less stressed, and more sales.
Not only can security cameras help manage rush hour by creating heat maps, they also can detect emotions as depicted by facial expressions. By playing ads to a selected group of people and scanning their reactions, marketers can save both time and money if the idea does not resonate enough with the target audience. This way of testing marketing is much faster and cheaper than traditional methods. It’s also more efficient and precise since it rules out the bias associated with surveys.
Social media platforms have already provided large quantities of raw data for social media listening. Until now, this was mainly text-based and focused on keywords or hashtags including brand or product names. The introduction of computer vision to the marketer’s toolset will extend this to image searching by a logo or another visual cue like the primary color, or even font.
The Store of the Future
We can see that such technology is already here just by looking at the Amazon Go experience. These brick and mortar stores are a perfect example of how computer vision can be used in offline environments. The scope is to create a seamless experience, through computer vision, sensors, and RFID tags.
When you enter the shop, you scan an app code which lets the system know you’ve started your shopping session. You take whatever you need and put in in your bag. When you leave the shop, the computer scans you for checkout. This gives a signal, and your card is charged much like after like an Uber ride. We can expect this kind of application at all physical locations, putting an end to shoplifting—and privacy.