Visual data recording and the IoT

IoT Devices Used

This story was delivered to BI Intelligence IoT Briefing subscribers. To learn more and subscribe, please click here.

Researchers at IBM recently announced that the firm's distributed visual data recognition models can now be trained faster than competing models from Facebook, according to TechCrunch.

Distributed visual data processing is a subset of deep learning, which is a branch of machine learning, that uses several graphics processing units (GPUs) to process and analyze massive data sets. It's ideal for very large deep learning projects where the data is too large to be processed on a single GPU.

While not specifically designed for the IoT, distributed visual data recognition could aid larger IoT ecosystems in particular.

Large IoT projects produce massive data sets that can't always be processed quickly and easily. Most computing systems and microchips aren't designed to process and analyze massive IoT projects involving hundreds of devices constantly producing data. That's especially problematic because the data IoT devices produce globally will rise to about 18 zetabytes (18 trillion gigabytes) as soon as next year, according to Cisco. The amount of data produced is so massive, in fact, that it often can't be transmitted to and from the cloud in time to keep the IoT ecosystem running smoothly.
Distributed visual data recognition could help alleviate some of the burden of analyzing all that data. Since distributed visual data recognition models can process massive data sets, they might also be able to handle the large data sets that are produced in many large IoT projects. But they likely won't revolutionize IoT data analytics — many IoT ecosystems, such as sensors on an oil rig, for instance, don't produce visual data at all — meaning the number of IoT ecosystems that could use distributed visual data recognition models might be limited to only a certain portion of IoT projects.

And that's crucial because large IoT projects are beginning to comprise a larger share of the overall IoT. In the past, many companies deployed only a few dozen IoT devices to enhance their operations, but that's starting to change as firms are scaling up their projects and may soon deploy hundreds of devices — 36% of global IoT projects now involve 100 or more devices. That means these firms will need to process much more data from these devices, which distributed visual data recognition models could help accomplish; this, in turn, could heighten demand for IBM's models in the coming years.

Peter Newman, research analyst for BI Intelligence, Business Insider's premium research service, has conducted an exclusive study with in-depth research into the field and created a detailed report on the IoT that:

Provides a primer on the basics of the IoT ecosystem
Offers forecasts for the IoT moving forward and highlights areas of interest in the coming years
Looks at who is and is not adopting the IoT, and why
Highlights drivers and challenges facing companies implementing IoT solutions

To get the full report, subscribe to an ALL-ACCESS Membership with BI Intelligence and gain immediate access to this report AND more than 250 other expertly researched deep-dive reports, subscriptions to all of our daily newsletters, and much more. >> Learn More Now

You can also purchase the report and download it immediately from our research store.

Join the conversation about this story »