Updated: Mar 25, 2020
Walking through the maze of rows in the many Halls at the Consumer Electronics Show (CES) 2020 in Las Vegas last week, it became readily apparent to even the most casual observer the importance of AI / ML to essentially every vertical market segment represented at the Show. CES is different. This is the one show, or maybe it is an event?, that attracts a worldwide audience of technologist and business leaders from a variety of key vertical market segments that includes a vast swath of technologies from the obvious: consumer electronics, to Transportation (both air and ground), Medical devices, Health and Fitness, Security, Home Automation, Robotics, Gaming and more. Across the board, isle after isle, manufacturers displayed their latest gadget and gizmo, most of which, included a machine learning capability used to detect, identify, or classify an object or person in order to effectuate a device function or feature. Although I was not completely surprised to see this first hand, I did pause to reflect on previous shows. It was not that long ago in which IoT was all the rage. I remember the industry luminaries pontificating on the IoT “drivers” that were going to bring to fruition the realization of the Internet of Things: low cost processor cores, low power devices, low cost sensors, and of course, lots of connectivity. All the connected edge devices were going to compose the outer layers of the vast networks that was going to drive Cloud computing. Why do I highlight this? It seems now as if the trend has somewhat reversed. Pushing data up into the cloud for processing is a great idea given the compute power of the latest and greatest servers. However, when one considers there could one day be billions or trillions of devices pushing data up, one concludes that much data could overwhelm the networks which today are designed more for data download as opposed to data upload.
Given the accelerating trend to add Intelligence to Edge Devices, more and more edge devices will be equipped not just with sensors but with cameras with the ability to generate vastly more data for analysis than ever before. The amount of data to be analyzed, processed and acted on will only continue to increase exponentially. This is coming at a time in which the requirements for edge devices are becoming more rigorous, not less. For many devices, there is a need for real time. And, there is a need for Security. Also, there is a need for Safety. The summation of all of this is the increased need for actual Intelligence at Edge as opposed to cloud based solutions. Devices will need to ability to detect, classify and act without relying on cloud computing.
Au-Zone has been leading the industry delivering vision and AI / ML solutions for embedded edge devices. Our experience in this domain has lead to the release of the Industry’s first commercial SDK for Machine Learning targeting low cost, general purpose embedded platforms : we call it DeepView.
Au-Zone has also announced the release of DeepView for Raspberry Pi. This is a no-cost trial version of DeepView designed for embedded developers to train, test, and deploy public or custom ML models on the Pi platform. This allows rapid exploration, proof of concept development, or product development prior to hardware selection or hardware availability. Stay tuned for details on where to download