Today, all of the visual recognition APIs are connected to the cloud making them not only extremely easy to deploy but highly scalable. However, not all data is equal and in some instances, this might not be the best route to take. Consider, for example, some of the intrinsic security issues of a cloud as listed by the CSA (Cloud Security Alliance): Data breaches, compromised credentials, account hijacking, malicious insiders, loss or theft of intellectual property, compliance violations or regulatory actions.
While Google Cloud, Amazon Web Services, IBM Bluemix or Microsoft Azure have proven resilient against these threats and offer very secure solutions for most needs, some data can and should never be exposed to even the risk, however low, of these threats.
What about those sensitive files?
Beyond security, there is also a heightened need to protect privacy. Some countries have very restrictive laws on where and how personal files, like photographs, can be shared and distributed. More than often, if it ever allows a cloud option, it limits the physical location of those servers within the same country as the company responsible for managing the data. Hard and costly to comply with a cloud option.
Visual content has always been a prime candidate for cloud distribution as they are, by nature, of a larger size, volume intensive and called upon via many devices and locations. But what is often omitted is how some of these files can be extremely sensitive and should not be handled carelessly. In law enforcement, a large consumer of very sensitive photos and videos, this data cannot and should not leave the premises. By extension, any institution or company involved in legal matters has the same, strict regulation.
More and more, countries have also passed laws to protect children and the circulation of their images. In Australia, for example, it is illegal to store or share images of underage children on a server that is not located in Australia. Even if it is to have them processed by a cloud based visual recognition service. Same goes for medical data, an absolute requirement for the protection of patients privacy.
So what are the options?
At Imagga, in order for everyone to benefit from visual recognition, the company has been hard at work to offer a choice: Cloud based or On-premise. Its On-premise solution, a first in the marketplace, offers the same high reliability and results as its award-winning cloud option.
It is hardware and OS agnostic, as long as the hardware and operating system respect the minimum specified requirement. Its processing power, the number of images that can be processed monthly, is only restricted by the size of the server. While its installation is a bit more challenging, both in networking skills and cost, it remains an extremely more affordable and efficient solution than building in-house.
Swisscom, a major provider of telecommunication solution, has elected to use Imagga’s On-premise solution to classify and auto tag its customers’ uploads of images. All the while fully respecting its customers’ privacy concerns, it is able to process millions of images seamlessly, all in the safety of its own managed servers. “Imagga impressed us with the quality of the recognition technology, recognizing both objects and broader scenic categories,” says Andreas Breitenmoser, Project Manager at Swisscom A.G.
Is the cloud dead?
Not at all. Each solution offers its advantages and disadvantages. It all depends on the data. Cloud offers a perfect option for companies that need deployment simplicity, secure processing, and easily scalable platform. It is perfect for those who need the full range of Visual A.I. with little deployment cost and manpower. On-Premise, however, is perfect for those with sensitive content and a very restrictive legislative environment. It offers the ultimate solution in security, privacy, and compliance.
Which one is good for me?
Figuring out which solution to use is not a complicated process. Considering time of deployment, cost of servers, range of visual A.I. solutions and particular requirements are factors to weight, but at a lesser priority. First, analyze your legislative environment. If your files cannot leave your servers, On Premise is the obvious and only solution. On the other hand, if your company already uses the cloud for other applications, especially for its visual content, then the answer is obvious: Cloud.
Author: Paul Melcher
Paul Melcher is the founder of Kaptur and Managing Director of Melcher System, a consultancy for visual technology firms. He is an entrepreneur, advisor, and consultant with a rich background in visual tech, content licensing, business strategy, and technology with more than 20 years experience in developing world-renowned photo-based companies with already two successful exits.