Powerful, AI-based OCR capabilities now available on the In-Sight 2800 Vision System
Cognex Corporation has expanded the capabilities of its In-Sight® 2800 vision system to include optical character recognition (OCR). The new ViDi EL Read tool, powered by edge learning technology, deciphers human-readable characters on reflective, low-contrast and non-flat surfaces with ease.
Edge learning technology simplifies character reading applications to improve quality control and traceability
“The In-Sight 2800 with edge learning redefines ease of use, allowing anyone to deploy AI-based applications regardless of experience level,” said Lavanya Manohar, Cognex Vice President of Vision Systems. “ViDi EL Read replaces complicated programming with example-based training, making it simple to set up OCR models to read characters on challenging surfaces and even multiple lines of text simultaneously.”
The easy-to-use OCR solution helps food and beverage producers to easily read expiration dates, even on curved surfaces, to verify freshness and prevent consumer recalls. Medical device and pharmaceutical companies can verify dates and lot codes to ensure vaccine efficacy and compliance with medical standards. Electric vehicle manufacturers can quickly locate and read the alphanumeric text laser etched on the bottom sides of EV batteries to improve traceability, and logistics facilities can decipher codes and text on a variety of package types to ensure proper routing and prevent rework.
OCR applications can be time-consuming to set up, requiring hours of programming by highly trained engineers, preventing many companies from automating this type of inspection. Using the ViDi EL Read tool, models are set up and deployed directly on the device in minutes with as few as 10 sample images. The accuracy of the results is measured by a confidence score that is displayed below each of the trained characters offering real-time, visual feedback and reassurance of reliable traceability across the supply chain.
The In-Sight 2800 vision system was first released in April 2022 designed to solve complex classification applications. Additional tools are currently in development and scheduled for release in coming months.