Todd Matsler, Director - Global Video Team, Intel Internet of Things Group
Ask a child to describe an image in a picture or a video, and chances are, they’ll be able to accurately describe what they see. Whether it’s a cat in a photograph or an airplane in a video, correctly identifying the content of digital image may be easy for a human, but ask a machine to recognize the same image, and the task becomes a bit more difficult – especially if context is required.
Whether it’s to make more intelligent and data-driven business decisions, speed discovery in life sciences, or keep our cities safer there are myriad uses for computer vision. Additionally, computer vision technology is advancing, becoming more accessible through easy-to-use tools and is quickly reshaping how we live and work. In fact, the computer vision market is expected to reach $20.84 billion by 2026, according to Reports and Data.
Additionally, with the proliferation of more and more IoT devices such as cameras, gateways and video analytics appliances that are producing data that needs to be analyzed close to the sensor capturing the action that’s occurring – to extract timely and useful information without unnecessary latency. That makes edge computing a key part of computer vision, from capturing and analyzing images to turning that data into something actionable.
The growth of artificial intelligence (AI) and specifically machine learning is enabling computer vision to become an increasingly powerful tool across all industries. From enabling the development of autonomous vehicles to spotting product defects on an assembly line to scanning x-rays to detect issues and improve medical outcomes.
AI isempowering businesses to use computer vision across multiple industries:
• In the healthcare industry, the combination of AI and IoT is streamlining drug discovery, speeding up genomics processing, and allowing medical imaging analysis for personalized treatment. In tests using the Intel OpenVINO toolkit, Philips developed bone-age-prediction and lung segmentationmodels. Philips achieved speed improvements of 188x improvement in bone-age-prediction and 38x speed improvement for thelung segmentationmodel and allowing them to deploy these solutions more cost effectively.
• In industrial manufacturing, IoT and AI are enabling smart manufacturing by using machine vision, connected devices, and real-time insights. A common product of aluminum casting is automotive parts, where quality assurance is critical.
Done correctly, using computer vision can unlock value from data, providing valuable insights in near real time that will help keep our cities safer, assist business decision makers in responding more quickly to customer needs, and helping medical professionals discover breakthroughs
Humans inspect for defects with their eyes. Factory workers must wait for the aluminum casting to cool down for inspection which is both dangerous and time-consuming. Including robotics and computer vision-based defect detection in the process alongside humans, YuMei’s defect detection capability increased five times from manual detection alone(including the time it took for parts to cool down), to automatic computer vision-basedinspection of freshly cast pieces.
• In retail, IoT and AI are solving critical problems like inventory distribution and enabling new experiences such as cashier-less checkout. Intel teamed with Pensa to enable AI and drones to autonomously scan shelves and alert retailers on inventory levels.
Steps to seeing the future with computer vision
While the notion of ‘computer vision’ may seem out of reach, perhaps a bit futuristic, and only used by large organizations, it doesn’t have to be. For companies looking to gain deeper insights, the steps to integrate computer vision to create more intelligent systems can be simple and well within reach. Here are foundational steps to determine if computer vision is right for you:
• Evaluating your need: It’s been estimated that half of the world’s data is being created in the physical world— from systems, sensors and tools inour cities, factories, medical facilities and beyond. Even if your organization doesn’t fall under one of those categories, chances are there’s a use case from your industry for using computer vision to gain actionable insights. Computer vision and machine learning may even help to reduce, or eliminate, shortcomings in current processes, including speed, cost, physical limits and potential for errors.
• Setting your sights: Processing data rich video and images by sending it back to the cloud is not only costly, it’s also time consuming. In the case of a self-driving car, where milliseconds matter, being able to process, store and analyze those images “at the edge” can save lives. Additionally, broadband connectivity and low-latency are critical for computer vision—but not always possible when sending data back to a data center for processing. Computer vision, often operating at the edge, can overcome the issues often associated by sending and processing data in the cloud. Determining latency and bandwidth requirements for your applications will help determine the right solution. It is important to ask the right questions:Do you need sub-second response times?Are the outputs, including images or videos, so large that bandwidth to the cloud would be cost-prohibitive?
• Partnering for success: The artificial intelligence and machine learning elements that make up computer vision can be extremely complex, but you don’t have to go it alone. Don’t allow lack of in-house skills or technology prevent you from exploring the benefits of computer vision. Intel, for example, provides developer-friendly tools for computer vision that are easy to use, plug-and-play, and scalable. On top of these tools, Intel has a broad network of systems integrators (SIs) and independent software vendors (ISVs), as well as an ecosystem of equipment and hardware providers that can meet all type of size, performance, and power requirements. Additionally, programs like Intel’s AI Builders program and Intel AI: In Production (AIIP) can help build and manage the right tools for any size enterprise.
As the number of connected devices continues to grow, so too will the need for computer vision. Done correctly, using computer vision can unlock value from data, providing valuable insights in near real time that will help keep our cities safer, assist business decision makers in responding more quickly to customer needs, and helping medical professionals discover breakthroughs.