IoT Forum meets in June at Sensors Expo in San Jose to look at innovation in Computer Vision. When coupled with the Internet of Things, computer vision allows devices to turn sensor data into actionable information. Join IoT Forum members, guests, and invited experts for an afternoon of innovation discovery, product demos, startup pitches, and great networking as we explore the opportunities around computer vision and machine vision in IoT.
On June 26, the IoT Forum held our meeting in San Jose, co-located with the Sensors Expo and Conference. Naturally, it made sense to focus our IoT meeting on the subject of sensors. For this year’s meeting at the Sensors Expo, our focus was on visual sensors and the sub-topic of Computer Vision.
We opened our session with an analysts presentation from Jeff Bier, a Founder at the Embedded Vision Alliance. Jeff had a few of solid observations that resonated throughout the rest of the presentations that came after. One was a defining vision for IoT, as the internet of things that interact with the physical world. To be IoT, a connected item must do one of both of: interact with the physical world via sensors that ingest data, or interact with the physical world via actuators that do something in the physical realm. Jeff defined Computer Vision as “extracting meaning from images and Video”. One other very interesting take-away from Jeff’s presentation was the (obvious only in retrospect) observation that, if you aggregate all sensor data that is generated in the world, data from image sensors is approximately 100% of all the data.
The good news is that image and video sensors are cheap and COTS equipment. Further good news is how flexible they are, as image data can be used to detect an incredibly high number of situations from motion, to vibration, to heartbeats, and much more. Of course, there is bad news: the costs of image and video data are in the data transport of the imagery, and in the analytics that must be used to extract useful information from millions of pixels.
Because of the high-bandwidth nature of the data flow from image sensors, a good portion of our panelists and speakers discussions hinged on the concept of Edge Processing. There is huge value in analyzing the video feeds at the source, and extracting the desired information there, such that the heavy data flow does not need to be transported back to a central cloud location.
Thanks to all who presented at our meeting. As always, IoT Forum Members can find presentations from this meeting in the Member Library.