Not just a pretty face in the crowd: The future of Visual Search
I’m fascinated with the potential for visual search a la Google Goggles. It’s one of the newest ways to search and at the forefront of the next generation: it allows you to search from your cell phone by snapping a picture, and returns information about the building, object, business, etc. (true augmented reality). I first used it when passing a historic building, and was curious about it. My friend pulled out his phone, snapped a picture, and voila! – information about what it was, the architect, date and style, etc. So neat that I think I actually squealed.
I’ve since used it again, with various success. It’s definitely still an emerging technology, but over time the database of images and capabilities will improve. Want more info on a product? Take a picture. Need info about a business? Photograph the storefront. Put simply, this thing packs some serious power, and its capabilities stretch far.
I personally think when it does improve (along with voice interactive software) it will become as indispensable to everyday life as cell phones, texting, and search engines have become.
But then I started thinking about eventual convergences, and the inevitable trajectory it will take: integration with facial recognition software and other data points.
Facial recognition software has had a huge influx of cash and interest since 9/11 for security reasons. It’s here, it’s improving, and pretty soon anonymity will be completely obsolete, if it’s not already – at least to the companies who use it to scan airport passengers, law enforcement, and others who have made it a goal.
We live in an era where an overwhelming amount of data exists on each of us, from our social networking connections and comments, to our shopping habits at the supermarket. Cell phone usage, online searches, cookies on sites visited, credit card purchases – all of these create data which builds a picture of who we are.
But currently, these are still siloed. The grocery store isn’t matching your checkout purchases with your Pandora list and identifying friends of yours on Twitter who are most likely to share your taste, then using the data to target them with advertising.
One of these days, though, facial recognition software will be one of the links connecting the dots between who you are with other data points such as your FB profile, and your Pandora list. At that point, if someone wants to know who that cute guy sitting at the next table is all they will have to do is take his picture – and know who he is, what music he likes, his address (courtesy of whitepages.com), books he’s bought (thanks to amazon.com), his house value (zillow.com), online subscriptions, health risks based on his grocery purchases, etc etc. Spokeo.com and a few others are baby steps towards data aggregation – crude, often incorrect, and using identifiers which are imprecise, but it is the next logical step in data mining: analysis crossing across collection points, as opposed to little ponds.
This scenario – inevitable as it is – obviously has many potential pitfalls. It’s great for companies (I’d advise anyone with a talent for numbers to consider a career in data modeling!), but is a mixed bag for consumers. The privacy issues are obvious, but those aside, the personalization that the market increasingly is demanding is impossible without data mining and developing good predictive capabilities. On the one hand people are uncomfortable with their data being gathered (not like this wasn’t always happening — it’s just more extensive now), and on the other, good data mining will ensure that people are targeted with offers and services that are interesting and relevant to them.
It’s a teetering tightrope walk. As a business strategist / consultant, I work with clients to develop strategies to take advantage of all that is legal, effective, and (personally) always try to do so with integrity. As consumers we should be trying to influence privacy legislation, to ensure that this future is one that not only makes our lives easier, but does so safely. The challenge is that data knows no national boundaries, so what effect will legislation be able to have? I don’t have the answer, only want to add to the discussion.