By Matt Pillar, Editor In Chief
After spending many hours cruising the aisles of the RILA and NRF LP conferences (May and June, respectively), it's clear to me that even the most astute analyst would be challenged to slice, dice, categorize, and differentiate the solution providers in the intelligent video space. It's a fervent marketplace right now; the vendors in it are like a cardboard box full of puppies, each very much the same, each very much different, all of them wiggling and pawing over one another as they try to establish their identities.
I asked Mark Holtenhoff for his perspective. His company, video analytics provider Aimetis, is effectively one of those puppies in the box, the one he thinks you should take home with you. Here are Holtenhoff's thoughts on the space and where it's headed.
Video is such a crowded and noisy marketplace. How are analytics providers attempting to differentiate themselves?
Video analytics is creating lots of excitement, lots of momentum and desire. But not all analytics are built the same. It's evolved into a competition of features. There are plenty of feature-rich solutions out there that are visual and exciting, but do they work? I think the only real way to test quality is to do side-by-side comparisons. We encourage tests and comparisons and pilots. It's only when retailers begin testing that many see past the features and become dismayed by the quality of the solutions out there.
You'll find that many intelligent video providers consider video analytics a separate entity. We don't see it that way. We think the two equally important elements of success are video analytics [turning raw video data into useful information] and video management [organizing that information in a way that makes it accessible and valuable to business]. We also don't believe these two elements should be handled separately. We believe that unstructured video data should be turned into useful information and extracted on the same platform.
What are the drivers of intelligent video adoption in retail?
It's typical for a retailer to spend several million dollars on video infrastructure and hardware. But once it's installed and running, what's it doing for them? Where's the business value? When retail execs realizes that for 5% of that hardware and infrastructure cost they can apply a video analytics layer that pulls real data and helps them make important business decisions, they tend to get excited and start thinking creatively. So creativity around the business opportunity is one driver. This is a competitive business with thin margins. Profitability doesn't come easily, and it's not attained by growth of the market; the market is mature. Driving market share is about competitive differentiators now, like higher conversions, increased efficiencies, and reduced shrink. Video has the potential to help retailers with all of these.
Another driver is the very prevalence of video in retail today. There's not a lot of intelligence just yet, but there are lots of cameras and DVRs and plenty of infrastructure. This ties into the business opportunity – there's generally more acceptance of a solution when its infrastructure has already been laid.
How are retailers applying intelligent video to the business across multiple disciplines?
We see six at least areas of opportunity, and they're not all tied directly to security. Intelligent, networked video takes the opportunity to nontraditional users. The six overarching opportunities we see for video in retail are:
Where will video play its next starring role in retail?
It's market driven. We're interested in learning about the challenge or business problem first, then delving into whether video offers an answer. Sometimes video can't help. Sometimes video can't help yet, not until there's a development.
We see opportunity for improvement in existing video applications. There is room for accuracy improvement and fewer false conclusions, for instance.
Apps that aren't proven yet but that we're working on include stereo analytics for 3-D video, which allows better consumer segmentation by size, age, and sex; color recognition, to either weed out or recognize employees in uniform, for instance; and minimizing the data footprint for analytics. Some retailers don't want to put a data center into the store, so we're exploring device-embedded apps and cloud computing models, for example.
If you're interested in retail applications for video analytics and would like to learn more, check out retailsolutionsonline.com or drop me a note at firstname.lastname@example.org.