It’s no secret that AI promises to transform industries and solve real-world problems.
“The power of AI has the potential to do all kinds of great things. It makes existing use cases and applications better, but it also opens up the door for new possibilities,” says Bill Pearson, VP of the Network and Edge Group, and General Manager of Solutions Engineering at Intel.
But while the benefits and opportunities are known, it’s not always clear how one can go about developing and deploying applications that meet expectations. Up until recently, AI has been limited to developers and data scientists with very specialized skill sets. And it hasn’t always been easy accessing the proper data to build and deploy models for AI applications.
“We’re finally at the point in the industry where we: one—have the data, two—have the computer power to process the data, and three—have the technology and tools to make it accessible for any business to use AI in unique, impactful ways,” says Pearson.
This enables all kinds of developers to start building edge AI applications for their business. But while it’s tempting to want to dive right in, developers should understand the foundation of AI as well as the tools and technologies available to them and then build from there.
AI Building Blocks, Get Started with Optical Character Recognition (OCR)
One of the early steppingstones is OCR. This is a very simple machine learning capability that has been around since 1965. It enables the ability to extract, convert, and repurpose data from documents and images through computer vision.
For instance, users can upload a document from an image or even upload a book and create an eBook. Or you may be more familiar with scanning and depositing a bank check from your mobile device. The postal service even uses OCR to automate the process of sending letters and packages, eliminating error-prone and time-consuming manual tasks.
Creating innovative and powerful #AI solutions goes beyond understanding the fundamentals. #Developers and businesses can leverage the right development tools in their toolbox to make it all possible. @Inteliot via @insightdottech
Enhance Businesses with Object Detection and Recognition
Going beyond OCR, object detection and recognition expands the capabilities and the problems that can be solved.
For instance, many shipping ports use OCR systems to automate container check-in, but then expand the solution to do other computer vision tasks like identifying free space and determining space utilization.
This real-time visibility is made possible with the AI capability known as object detection, which is one of the most commonly known AI applications today because it allows users to do things like:
- Defect detection on the manufacturing line
- Predictive maintenance on equipment
- Inventory management in warehouses
- Weed detection in agriculture fields
With object detection and recognition, systems can also prevent car accidents before they happen and even alert pedestrians of oncoming traffic. Medical professionals use object detection to provide better care and accurate diagnosis to patients. “This idea of taking images—and a lot of medical technology today relies on imaging—and being able to leverage AI to help identify and detect disease early on is really powerful,” says Pearson.
The food service industry is beginning to use AI and object detection to improve the quality of service for customers. For instance, PreciTaste, a smart software automation provider, uses edge AI and object detection to improve the quality of service for customers. The company does this by not only leveraging images but also video data. The ability to find and categorize activities within a recorded or live video is known as human action recognition.
PreciTaste’s software can leverage video feeds to detect whether a quick-service restaurant worker properly packaged a takeout order and alert them of any potential errors. It also observes restaurant inventory and predicts customer demand throughout the day to help match their production based on that information. “There’s major demand for reliability, which is fantastic to see. Not to mention it has its own benefits. One being less waste—users can more accurately determine what level of resources or ingredients they need to meet their demand. It also makes for faster service for customers. Who doesn’t like getting their meal quickly?” says Pearson.
Vision recognition applications also extend to other industries such as smart cities for safety and security as well as manufacturing for monitoring the factory floor. For example, Vulcan AI, an AI workplace solution provider, has a solution called WorkSafe, where they leverage their workplace cameras to analyze video data and identify safety hazards. “Now manufacturers can flag safety hazards very, very quickly—allowing them to make major improvements,” says Pearson.
Empower AI Developers with the Right Tools
Creating innovative and powerful AI solutions goes beyond understanding the fundamentals. Developers and businesses can leverage the right development tools in their toolbox to make it all possible—and with the right software, they can speed up their edge AI development.
For instance, the Intel® Edge Software Hub allows developers to leverage prevalidated software, including use cases and reference implementations based on real-world applications, to experiment, test, create, and optimize their edge AI solutions.
The OpenVINO™ Toolkit becomes a powerful tool for edge AI developers. Because of its “write once, play anywhere” approach, they can create an application or algorithm once and deploy it across a variety of hardware architectures, according to Pearson. For a curated OpenVINO experience, developers can take advantage of the Intel® Developer Cloud for the Edge, which allows them to evaluate, benchmark, and protype their AI and edge solutions on Intel® hardware—no matter where they are in their development process.
“Once you have a new capability like AI, and you put it in the hands of developers, they’re going to add their magic to it and do fascinating things that we never thought of,” Pearson says.
But they can’t create every solution alone. In manufacturing, for example, developers can create a defect detection solution, but they may not be the subject matter experts in this area, and often need to leverage the expertise of the operations team to create a solution that solves a problem.
To make it easier to bring domain users, subject matter experts, and business users into the AI development lifecycle, Intel offers the Intel® Geti™ Platform.
“We’ve been on a journey to make a developer’s life easier, to give them more tools, more foundational software, and more capability to figure out the problems they’re trying to solve more effectively and efficiently,” says Pearson.
Intel Geti is designed to simplify the AI training model process by enabling collaboration with developers and subject matter experts to label, train, optimize, and deploy computer vision models, according to Pearson. “Now, developers and non-developers can come into this easy-to-use interface, identify the images with defects, identify the good images, and very, very quickly build a model that’s been trained, modified and updated,” he says. “It simplifies labor-intensive tasks, ultimately freeing those individuals up to focus on new areas.”
Intel Geti also integrates with OpenVINO so once the domain experts or business users develop high-quality models, developers can leverage the AI toolkit to deploy them into the real world.
Beyond Computer Vision
As you can see, AI is solving all kinds of real-world problems, but computer vision is just the start. There are also many AI-based audio and speech solutions and opportunities out there today.
For example, speech-to-text recognition has become an important part of our daily lives. It’s helping drivers safely read and send messages while they are on the road. AI-based audio solutions are also helping to remove or reduce background noises in phone calls and virtual meetings, which is extremely important in today’s hybrid and remote workplace.
“We’re using technology to solve real-world challenges and to help make our livelihoods safer. It’s the heart of the work developers are doing that makes these solutions a reality. Give them the right tools, and they’ll develop the use cases, models, and apply the technology in meaningful ways,” says Pearson. “When tools can bring AI development and the business together, that’s when we really get life-changing, world-changing results.”