“While AI promises to revolutionise quality inspection, integrators and designers struggle with how to cost-effectively integrate advanced capabilities into existing and new applications,” explained Harry Page, President, Pleora Technologies. “Our AI Gateway platform is an evolutionary approach to AI, with plug-in machine learning skills for classification, sorting, and defect detection, combined with the flexibility to train and deploy open source or custom algorithms. Users can immediately employ AI to reduce costly inspection errors, false-positives, and secondary screenings while preparing for more advanced Internet of Things (IoT) and Industry 4.0 applications.”
Pleora’s gateway will enable end-users and integrators to deploy AI skills without any additional programming knowledge. It uses a simple web-based interface, while images and data are uploaded to “no code” training software on a host PC, which generates a neural network that is deployed onto the Pleora AI Gateway.
For applications requiring unique AI capabilities, an operating system – built on Pleora’s eBUS SDK – provides a framework to upload custom skills developed in Python to the gateway. Pleora’s AI Gateway then automatically handles image acquisition from the camera source and sending out the processed data over GigE Vision. The gateway operating system supports development around popular open source frameworks like TensorFlow and OpenCV and leverages the built-in NVIDIA GPU for hardware acceleration.
The Pleora AI Gateway is designed to work with existing inspection hardware and software. The gateway interfaces with GigE Vision, USB3 Vision, Camera Link, and MIPI cameras from any vendor, allowing designers to retain existing infrastructure while adding AI skills to their process. The Pleora gateway receives the camera feed, automatically performs the deployed plug-in or custom AI skill, and transmits the pre-processed data over a real-time GigE Vision connection to the existing inspection application.
The configurable platform is built on an NVIDIA GPU that is easily upgraded for applications requiring more powerful AI image processing, and multiple gateways can be networked to enable distributed image processing to leverage the multicast capabilities of GigE Vision.
