Camera manufacturer PTZOptics (Downington, PA, USA), AI-software developer Detect-It (Oak Park, MI, USA), and computer hardware supplier Comtronix (Young Harris, GA, USA) have developed a platform for AI inspections. The system is designed to enable manufacturers, including automotive, heavy equipment, and food and beverage producers, to deploy on-site real-time inspection, automation, and intelligence cost-effectively and easily.
One goal of the collaboration is to demonstrate the value, ease, and cost-effectiveness of on-premises AI deployment, according to PTZOptics, Detect-It, and Comtronix officials. Unlike cloud-dependent AI solutions, this system runs entirely behind a customer’s security system. That means video feeds stay inside the plant, operational data is never transmitted externally, and the system remains fully functional even without internet connectivity.
“Our customers require real-time, multi-camera intelligence that delivers actionable insights instantly—even in the harshest environments,” Comtronix Vice President Jimmy Haugen said. “They expect systems to handle complex deep learning models for advanced quality control, defect detection, safety monitoring, and robotic guidance, all while maintaining high reliability, low power consumption, and seamless integration with existing OT infrastructure.”
When it comes to deployment of inspection systems, there are generally two architectural approaches they tend to see customers using, a “local approach,” in which a single or multiple PCs are used right on the factory floor hard-wired to cameras in the station, and a “network approach, in which PCs are stored in a cold room or IT closet and connected via video LAN to cameras deployed on the factory floor, Detect-It COO/CMO Rick DiLoreto said.
“Simple inspections can be run by a bookshelf PC and will support 3-5 cameras running multiple nets,” DiLoreto said. “The more cameras you add the more horsepower you will need. To scale up the system just add additional PCs, cameras, and Net Runner licenses.”
“Manufacturers are under pressure to improve quality, reduce defects, and bring AI into production, but many teams still see AI as expensive, complicated, and difficult to maintain,” said Matt Davis, CTO, PTZOptics. “This collaboration shows a more practical path.”
Cheap cameras, designed for consumer grade vision or security applications alone can fail under harsh industrial conditions, Davis said. But with high-quality robotic cameras, visual AI software, and industrial computers working together, manufacturers can easily start with a simple inspection use case and rapidly scale toward more advanced, factory-wide automation.
Detect-It’s AI software is designed to enable teams to train visual detections from video clips, rather than single still images, creating robust models and help manufacturers inspect parts, verify assembly steps, identify defects, and trigger workflow actions quickly and efficiently, DiLoreto said.
Comtronix will provide the industrial computing layer for the demonstrations, including AI and machine learning PCs and servers designed for demanding industrial environments.
“Manufacturing AI needs more than a model, it needs hardware that can survive and scale in factory floor conditions,” said Jimmy Haugen, vice president of Comtronix. “Our industrial computing systems are built for these environments, with the processing power and expandability customers need as they move from their first inspection station to larger and more complex AI deployments.”
Related: How to Build an AI-Enabled Industrial Vision System
