HEALTHCARE

AI and computer vision have become commonplace in manufacturing. In manufacturing, if you can see the data, there’s something you can do with the data. But not every industry has that mindset, or the luxury of data scientists on-site. That shouldn’t stop the many new and exciting use cases for AI—from medicine to traffic to agriculture—from taking advantage of these tools.

We talk with Elizabeth Spears, Co-Founder and Chief Product Officer at Plainsight (formerly known as Sixgill), a machine learning lifecycle management provider for AIoT platforms, and with Bridget Martin, Director of Industrial AI and Analytics of the Internet of Things Group at Intel^®, about the accessibility and democratization of AI, and how these factors are key to getting the most out of this crucial technology—for companies and, ultimately, for consumers.

Where do things stand right now in terms of new applications in the manufacturing space?

Bridget Martin: There are two different perspectives. Some manufacturers we would consider more mature—where there are automated compute machines already on the factory floor, or in individual processes on the manufacturing floor. They are automating processes, but also—and this is critical when we’re talking about AI—outputting data. And that could be the metadata of the sensors, or of the processes that that automated tool is performing.

These manufacturers are really looking to take advantage of the data that’s already being generated in order to predict and avoid unplanned downtime for those automated tools. This is where we’re seeing an increase in predictive maintenance–type applications and usages.

But then you also have a significant portion of the world that is still doing a lot of manual manufacturing applications. And those less mature markets want to skip some of the automation phases by leveraging computer vision—deploying cameras to identify opportunities to improve their overall factory production, as well as the workflow of the widgets going through the supply chain within their factories.

Can you talk about some new applications making use of this technology?

Elizabeth Spears: A really cool one, which is just becoming possible, is super resolution. One of the places where they’re researching its application is in using less radiation in CT scans. Think of one of those FBI investigation movies where they’re looking for a suspect, and there’s some grainy image of a license plate or a person’s face. And the investigator says, “Enhance that image.” All of a sudden it becomes this sharp image, and they know who did the crime. That technology really does exist now.

Another example is in simulating environments for training purposes in cases where the data itself is hard to get. Think about car crashes, or gun detection. In those cases, you want your models to be really accurate, but it’s hard to get data to train your models with. So just like in a video game, where you have a simulated environment, you can do the same thing to create data. Companies like Tesla are using this for crash detection.

It’s really across industries, and there’s so much low-hanging fruit where you can really build on quick wins. My favorite cases around computer vision are the really practical ones. And they can be small cases, but they provide really high value.

One that we’ve worked on is just counting cattle accurately—and that represents tens of millions of dollars in savings for that company.

How can organizations recognize their AI use cases and leverage computer vision?

Elizabeth Spears: I feel like we often talk about AI as though an organization has to go through a huge transformation to take advantage of it, and it has to be this gigantic investment of time and money. But what we find is that when you can implement solutions in weeks, you can get these quick wins. And that is really what starts to build value.

For us it’s really about expanding AI through accessibility—AI isn’t just for the top-five largest companies in the world. And we want to make it accessible not just through simplified tools but also simplified best practices. When you can bake some of those best practices into the platform itself, companies can have a lot more confidence using the technology. We do a lot of education in our conversations with customers, and we talk to a lot of different departments; we’re not just talking to data scientists. We like to really dig into what our customers need, and then be able to talk through how the technology can be applied.

“For us it’s really about expanding #AI through #accessibility—AI isn’t just for the top-five largest companies in the world.” – Elizabeth Spears, Co-Founder and Chief Product Officer @plainsightAI via @insightdottech

Hiring machine learning and data science talent is really difficult right now. And even if you do have those big teams, building out an end-to-end platform to be able to build these models, train them, monitor them, deploy them, keep them up to date, and provide the continuous training that many of these models require to stay accurate—that all requires a lot of different types of engineers.

So, it’s a huge undertaking—if you don’t have a tool for it. That’s why we built this platform end-to-end, so that it would be more accessible and simpler for organizations to be able to just adopt it.

What are some of the challenges to democratizing AI and what is Intel^® doing to address those?

Bridget Martin: Complexity is absolutely the biggest barrier to adoption. As Elizabeth mentioned, data scientists are few and far between, and they’re extremely expensive in most cases. This concept of democratizing AI and enabling, say, the farmers themselves to create these AI-training pipelines and models, and to deploy, retrain, and keep them up to date—that’s going to be the holy grail for this technology.

We’re talking about really putting these tools in the hands of subject-matter experts. It gets us out of the old cycle—take a quality-inspection use case—where you have a factory operator who would typically be manually inspecting each of the parts going through the system. When you automate that type of scenario, typically that factory operator needs to be in constant communication with the data scientist who is developing the model so that the data scientist can ensure that the data they’re using to train their model is labeled correctly.

Now, what if you’re able to remove multiple steps from that process and enable that factory operator or that subject-matter expert to label that data themselves—give them the ability to create a training pipeline themselves. It sounds like a crazy idea—enabling non–data scientists to have that function—but that’s exactly the kind of tooling that we need in order to actually properly democratize AI.

Because when you start to put these tools in people’s hands, and they start to think of new, creative ways to apply those tools to build new things—that’s when we’re really going to see a significant explosion of AI technologies. We’re going to start to see use cases that I, or Elizabeth, or the plethora of data scientists out there, have never thought about before.

Intel is doing a multitude of things in this space to enable deployment into unique scenarios and to lower the complexity. For example, with Intel^® Edge Insights for Industrial we help stitch together an end-to-end pipeline as well as provide a blueprint for how users can create these solutions. We also have configuration-deployment tools to help system integrators install technology. For example, if a SI is installing a camera, our tools can help determine the best resolution and lightning. All these factors have a great impact on the ability to train and deploy AI pipelines and models.

How can organizations go about starting this journey?

Elizabeth Spears: There are so many great resources on the internet now—courses and webinars and things like that. There’s a whole learning section on the Plainsight website, and we do a lot of “intro to computer vision” events for beginners.

But we also we have events for experts—where they can find out how to use the platform, and how to speed up their process and have more reliable deployments. We really like being partners with our customers. So, we research what they’re working on, and we find other products that might apply as well. We like really taking them from idea, all the way to a solution that’s production ready and really works for their organization. 

How is Intel working through its ecosystem to enable its partners, end users, and customers?

Bridget Martin: One of my favorite ways of approaching this is to really partner with that end customer to understand what they’re ultimately trying to achieve, and then work backward. Also, one of the great things about AI is that you don’t have to take down your entire manufacturing process in order to start playing with it. It’s relatively easy to deploy a camera and some lighting and point it at a tool or a process. And so that is really going to be one of the best ways to get started.

And of course, we have all kinds of ecosystem partners and players that we can recommend to the end customers—partners who really specialize in the different areas that the customer is either wanting to get to, or that they’re experiencing some pain points in.

How does Plainsight address scalability, and how does Intel help make an impact here?

Elizabeth Spears: We look at scale from the start, because our customers have big use cases with a lot of data. But another way you can look at it is to scale through the organization, which really comes back to educating more people. We’ll talk to a specific department within a company, and someone will say, “I have a colleague in this other department that has a different problem. Would it work for that?”

Concerning Intel—because we’re a software solution, Intel’s hardware is definitely one of the places that we utilize them. But they’re also really amazing with their partners—bringing partners together to give enterprises great solutions. 

What do you both see as some of the most exciting emerging opportunities for computer vision?

Bridget Martin: One, I would say, is actually that concept of scalability. Not just scaling to different use cases, but also scaling to different hardware—there’s no realistic scenario where there is just one type of compute device involved. I think that’s going to be extremely influential, and really help transform the different industries that are going to be leveraging AI.

But what’s really exciting is this move toward democratization of AI—really enabling people who don’t necessarily have a PhD or specialized education in AI or machine learning to take advantage of that technology.

Elizabeth Spears: I agree. Getting accessible tools into the hands of subject-matter experts and end users, making it really simple to implement solutions quickly, and then being able to expand on that. It’s less about really big AI transformations, and more about identifying all of these smaller use cases or building blocks that you can start doing really quickly, that over time make a really big difference in a business.

Related Content

To learn more about the future of democratizing AI, listen to Democratizing AI for All with Plainsight and Intel^® and read Build ML Models with a No-Code Platform. For the latest innovations from Plainsight, follow them on Twitter at @PlainsightAI and on LinkedIn at Plainsight.

This article was edited by Christina Cardoza, Senior Editor for insight.tech.