Representative End User Clients
Representative Automation Clients
Representative Software Clients
In a recent episode of the ARC Advisory Group's Digital Transformation podcast, Colin Masson, Research Director for Industrial AI at ARC, engaged in a thought-provoking conversation with his colleague, Larry O'Brien, about the transformative power of computer vision within the industrial AI landscape. Their discussion illuminated key use cases, explored the complex market dynamics, and provided actionable insights for industrial organizations eager to leverage this cutting-edge technology.
To delve deeper into this fascinating topic and hear the full conversation, we encourage you to listen to the full podcast episode on the ARC Advisory Group's Digital Transformation podcast channel on Buzzsprout.
You can also watch easily digestible selected SPARCs (Short Podcasts by ARC) on YouTube for a quick overview of the topics discussed.
Defining Computer Vision: Beyond Machine Vision
Larry O'Brien began by clarifying the fundamental concept of computer vision, stating, "That's really any kind of visual input that's going to go into a computer." He distinguished it from machine vision, explaining, "A lot of people use the term machine vision. I personally think that's a little more limited to inspection applications and things coming off an assembly line… it's really just computerizing any kind of video or optical kind of input". This broad definition encompasses various inputs, including closed-circuit TV, infrared cameras, and other sensing technologies, highlighting the pervasiveness of computer vision in today's industrial environments.
A Spectrum of Use Cases: From Safety to Quality
Colin and Larry highlight a diverse range of applications for computer vision in the industrial sector, emphasizing its versatility and potential impact. These use cases include:
Worker Safety: Ensuring workers are wearing proper Personal Protective Equipment (PPE) in designated areas, addressing a critical aspect of industrial operations.
Security and Surveillance: Implementing robust perimeter protection, monitoring for unauthorized access, and enhancing overall site security using CCTV. As Larry noted, "You want to make sure you have a full, 360-degree view of what's happening."
Remote Inspection and Monitoring: Utilizing robots and drones to inspect remote assets like high-voltage transmission lines or sensitive areas of a plant, as well as difficult to reach or hazardous areas. This enhances both safety and efficiency.
Quality Inspection: Conducting high-speed, real-time inspections of products on assembly lines to identify defects and irregularities, ensuring high standards of manufacturing quality.
Anomaly Detection: Employing machine learning to detect deviations from normal patterns, indicative of potential maintenance issues, contributing to predictive maintenance strategies. Colin pointed out that machine learning for anomaly detection is "by far the most pervasive" application of AI, closely followed by quality inspection using computer vision.
Employee Health and Safety: Leveraging thermal imaging cameras to monitor employee health and ensure a safe working environment.
Larry emphasized the importance of the pattern recognition capabilities of AI, noting that it allows systems to "go through huge amounts of imagery and be able to pick out that one thing that maybe you need to work on, or that one asset that's about to fail.”
Navigating the Complex Supplier Landscape
The discussion also delved into the complex and rapidly evolving supplier landscape that has emerged in response to the increasing adoption of computer vision. Larry identified five key classes of suppliers:
Traditional Machine Vision Providers: These suppliers are evolving by incorporating AI to enhance product functionality and user experience, often focusing on inspection-related applications.
Hyperscalers: Cloud service providers like AWS, Microsoft, and Google offer broad AI vision capabilities via APIs, requiring systems integrators to create complete, tailored solutions. As Larry noted, "The Hyperscalers aren’t going to provide you with a packaged industrial solution right, right out of the box”.
CCTV and Video Surveillance Providers: Traditional security-focused companies, such as Bosch and Honeywell, are expanding their offerings to include employee health and safety applications, integrating multiple functions into a single, IoT-enabled environment.
Robotics and Drone Inspection Providers: Companies like Boston Dynamics and SP Robotic Works, are embedding computer vision capabilities into their hardware, but often limited to their own specific robotics or drone systems.
Open Foundations and Consortia: These initiatives provide open-source tools and libraries, like OpenCV (Open Source Computer Vision Library), TensorFlow, and others, for organizations interested in developing customized computer vision solutions.
Larry highlighted that this varied landscape creates a lot of change and volatility, saying, “Basically, we have about five classes of suppliers that are all converging on this space at once—which is typical for these forming markets. There's a kind of chaos, and a lot of volatility going on right now." He also emphasized the critical importance of considering a supplier’s security posture, adding “If you don't have several key cyber selection related selection criteria, you're obviously making a huge mistake.”
Key Takeaways and Actionable Insights
Colin and Larry's conversation provided several key takeaways and actionable insights for industrial organizations:
Computer vision is already present, the focus should be on extracting value: As Larry succinctly put it, "you've already adopted computer vision... the question is whether or not you have applied the right analytics and AI to this computer vision to make better decisions about your business?"
A unified approach to computer vision yields significant benefits: A unified approach to applying AI and analytics across all computer vision applications can lead to substantial cost reductions, safer operations, and improved data gathering. Larry noted, "Some of the users we've spoken to say they expect 10-15 percent reductions in security and monitoring expenses."
Careful supplier selection is paramount: Organizations must carefully evaluate suppliers based on their specific needs, capabilities, long-term stability, and service expertise. Larry cautioned against overly customized solutions, stating, "I would caution anybody that would be doing that, because it could become hard to maintain" He also emphasized that "many of the suppliers in this space tend not to have extensive, substantial service capabilities."
Start with the low-hanging fruit: Larry noted that implementing AI based computer vision solutions is a great way to start a digital transformation program because "It's a relatively easy way to take a lot of this video input and apply AI and get a benefit from it quickly and then expand your digital transformation."
A Note of Caution and a Call to Action
While the potential of computer vision in industrial AI is immense, both Colin and Larry emphasized the importance of careful planning and execution. Larry cautioned against excessive customization and urged users to evaluate the long-term viability of their chosen vendors.
Interest SPARCed?
Thanks for tuning in to this SPARC! Share these episodes with others who are passionate about applying technology the right way to address skills gaps, create more intelligent processes, while building a more profitable and sustainable future. For those looking to go deeper, ARC Advisory Group offers longer format Digital Transformation and Sustainability Podcasts, and unparalleled guidance in developing digital transformation and sustainability strategies for industrial organizations. Reach out to us for expert insights and support that can help you turn these ideas into action and lead your organization toward a more profitable and sustainable path.
To contribute to SPARCs, or the Digital Transformation and Sustainability Podcasts, contact Colin Masson and Jim Frazer at ARC Advisory Group through your client manager.
For ARC Advisory Group recommendations for closing the digital divide by embracing Industrial AI, and governing and guiding major decisions about enterprise, cloud, industrial edge and AI software, please contact Colin Masson at [email protected].