When Thomas Edison and his colleagues formed General Electric in 1892, they began harnessing the power of electricity to transform everything from lighting and transportation to manufacturing and medical equipment. Today, as we discussed during the re:MARS conference in Las Vegas this week, GE Healthcare, Intel, and a number of other ecosystem partners are at the forefront of a new age of democratization that is taking artificial intelligence (AI) from prototype to production.
AI is driving a digital transformation across industry sectors that is every bit as powerful and profound as electricity. Healthcare, retail, industrial, automotive, robotics, and smart cities applications are already putting AI to work. What’s ahead amounts to an artificial intelligence factory for everyone. The confluence of deep learning – which uses algorithms based on neural networks similar to the human brain – edge processing, and the new 5G wireless standard will together handle the massive amounts of data powering the Internet of Things (IoT).
Connected devices, fueled by AI, are going to unlock the potential for machines and humans to ingest, curate, train, and deploy data for improving our health, making our work environments safer, and improving the efficiency of our cities. GE Healthcare, for example, is already using AI at scale in medical imaging diagnostics and monitoring equipment. That means radiologists now have a powerful tool to more accurately and more quickly identify and diagnose conditions such as pneumothorax, commonly known as a collapsed lung.
GE Healthcare’s devices use a deep learning framework, part of its Edison AI platform, to automate the process of obtaining and analyzing each image in order to expedite the review of critical findings. The AI-assisted devices are faster, more accurate, and more efficient.
OpenVINO™ Toolkit: The Free AI SDK
GE Healthcare is using Intel processors and the Intel® Distribution of OpenVINO™ toolkit to implement AI algorithms for enhancing imaging. The OpenVINO toolkit, Intel’s visual inference and neural network optimization software development kit (SDK), is designed to extend workloads and optimize performance across Intel® Vision Products, a comprehensive portfolio of AI-enabling processors and accelerators.
GE Healthcare, for example, has optimized its AI algorithms to scan X-ray imagery, reducing the time it takes to detect a pneumothorax to within seconds at the point of care. The pneumothorax inferencing models will potentially help radiologists achieve better diagnostics and outcomes.
“Delivering higher quality care at a lower cost is imperative to deliver on the promise of precision health, and it depends upon the pervasive use of AI,” said Keith Bigelow, General Manager & Senior Vice President AI & Analytics, GE Healthcare. “Our Edison intelligence platform uses the OpenVINO toolkit API to deploy algorithms directly on our medical devices – harnessing the computing power of the Intel CPU that is already present and delivering sub-second inferencing times. Intel’s mission to ‘make AI at the edge’ as free and easy as it is to write software matches our vision to democratize AI in healthcare.”
Intel’s deep learning SDK is well-suited for inferencing applications across industry sectors because it is optimized to facilitate high performance processing and the effectiveness of computer vision applications running on Intel hardware. Another significant benefit of OpenVINO toolkit is that AI customers who are already using Intel processors do not have to invest in additional hardware upgrades in order to implement deep learning solutions.
The functionality of the OpenVINO toolkit, which just marked its first anniversary, now includes the first-ever fully automated computer vision annotation tool. That means Intel’s free SDK for tasks such as interpolation of annotations and semantic segmentation now has the unique ability to handle inference-based automatic annotation of videos.
The Evolution of Robots
The OpenVINO toolkit is also going to be one of the driving forces behind adding visual intelligence to robotics. Not only is the OpenVINO toolkit powerful and easy-to-use, it also helps prevent hardware overheating – a significant concern for robot developers – by managing the allocation of resources between CPU/GPU and VPU.
Intel has collaborated with Amazon Web Services and Aaeon, which is rolling-out its UP Squared RoboMaker Developer Kit later this month. AAEON’s UP Squared series uses powerful Intel processors, vision accelerators, and RealSense cameras that give developers a simple way to add proven AI capabilities to their edge computing applications. AWS RoboMaker is a service that makes it easy to develop, test, and deploy intelligent robotics applications at scale.
“We are excited to be collaborating with AAEON and Intel to enable developers to easily build, simulate, test, and deploy intelligent robotic applications using AWS RoboMaker,” said Roger Barga, General Manager of AWS Robotics and Automation Services.
Robotics is at the heart of the fourth industrial revolution, Industry 4.0. Autonomous robots are increasingly capable of guiding products from ordering to production to delivery. ABI research projects that the compound annual growth rate for commercial robots will be 63.2 percent between 2017 and 2027.
Smart Cities Are Here
As the IoT continues to evolve, one of the most promising applications for improving our daily lives is smart cities, which will integrate AI into the fabric of everyday life. Smart city technology has a voracious appetite for data, and AI — combined with edge processing power — is enabling its development. Intel’s hardware and software solutions are helping get data out of silos and putting it to work for everything from smart lighting and mobility to computer vision and edge-to-cloud connectivity.
Half of the world’s people currently live in cities, and by 2030 that figure will increase to 60 percent, according to projections from the United Nations. Intel is at the center of innovation for cities of today and the future—powering every segment of the connected world. Smart city mobility solutions, for example, include smart parking, traffic management, and public transit optimization. In one study, we found that AI-driven traffic management systems could save commuters up to three weeks of time per year.
Mining the vast volumes of data generated by IoT devices promises to deliver improved business insights, efficiency, and productivity. Intel and its IoT partner ecosystem are leading the way with edge-to-cloud solutions optimized for data-intensive workloads across industry sectors. From smart cities to smart factories, AI is now the common denominator for collaboration among machines, humans, and enterprise systems.
For more information, https://www.intel.ai/remars/