Headwall & perClass Partner to Advance Embedded Vision Solutions with Intuitive Spectral Imaging

Industry-leading platform paired with real-time onboard classification on processing/production lines

Headwall Photonics, a world leader in hyperspectral imaging platforms and instrumentation, and perClass BV, provider of innovative and reliable software for production deployment of machine learning and spectral imaging solutions, announced a partnership that will help meet challenging needs for customers worldwide.

“Markets such as food processing, high-tech manufacturing, and material-sorting for recycling now have a much easier path to exploit the power of hyperspectral imaging,” says Christian Felsheim, Director of Headwall Photonics Europe. “Headwall’s new Hyperspec MV.X embedded hyperspectral imaging platform is designed from the start to run spectral classification algorithms such as those developed by perClass Mira and output actionable data in real time, enabling automated sorting systems to take immediate action along the line.”

Recommended AI News: Lynx Analytics Releases Lynxkite 4.0, An Open Source Graph Data Science Platform

Dr. Pavel Paclik, Founder and Managing Director of perClass BV, agrees and adds, “Our mission at perClass is to significantly lower the barrier to entry for the vast number of industries that can benefit from an AI-based workflow that creates highly accurate models without the need for cumbersome coding. We’re excited to be working with Headwall’s architecture so that customers can fully benefit from the many advantages our own automated machine-learning platform offers.”

Recommended AI News: NetFoundry And Cloud Solution Partners Extend Zero Trust Networking To Google Cloud And Anthos Applications

Accurate and reliable classification models can be generated using the Headwall MV.X and perClass Mira software on a PC. Those models are then uploaded to IP67-rated MV.X imaging systems on the production line with no need for dedicated computers to store and process the raw hyperspectral data. Amazingly the compact MV.X systems host a runtime version of perClass Mira that outputs the classified data through an industry-standard GenICam interface to robotic systems to take action along the line without any further processing.

The pairing of Headwall’s MV.X and perClass’ Mira platforms allow customers to rapidly deploy solutions that can be easily updated if requirements or situations change; for example, the material being inspected changes or even more accurate classification models are developed over time.