The machine vision software manufacturer MVTec Software has announced a technology collaboration with Qualcomm Technologies to enable customers of MVTec HALCON software to access Qualcomm Dragonwing processors’ high-performance NPU (Neural Processing Units). This access particularly benefits users of smart cameras, allowing Deep Learning applications to run on embedded devices with considerably higher speed and efficiency. As an initial step, MVTec is developing an interface between MVTec HALCON software and the Qualcomm Dragonwing RB3 Gen 2, leveraging Qualcomm Technologies’ upstream Linux and AI framework.
Making deep learning even more usable
“As a pure software manufacturer for machine vision, our goal is to ensure that customers can develop the best possible machine vision solution using our products. However, this is only possible when combined with the most powerful hardware available. That’s why we place great emphasis on technology collaborations. With its powerful AI accelerator, Qualcomm Technologies is an ideal partner to meet the requirements of machine vision. We are therefore very pleased to welcome this technology to our Technology Partner Program,” explains Roman Moie, Product Owner at MVTec.
Eric Mazzoleni, Vice President, Industrial and Embedded IoT Sales – Europe at Qualcomm Germany, adds: “Industrial image processing is evolving rapidly with advancements in AI. By deploying MVTec’s software directly on the dedicated NPU, the Dragonwing processor unlocks new levels of performance for demanding machine vision workloads. Together, we’re helping manufacturers meet the requirements of Industry 4.0 for Europe’s industrial transformation.”
MVTec offers the powerful standard software HALCON, the low-code software MERLIC, and the Deep Learning Tool for labelling image data. MVTec’s software products are used in nearly all industrial sectors and across a wide range of applications. The Dragonwing RB3 Gen 2 is a powerful platform powered by the Dragonwing QCS6490 processor. With the Qualcomm Hexagon fused AI-accelerator, it delivers up to 12 dense TOPs and enables faster and more power-efficient AI inferencing for industrial machine vision applications.
Higher speeds for deep learning applications on smart cameras
With Embedded Vision, MVTec offers software solutions specifically tailored to the requirements of embedded vision devices. The key objective is that MVTec software products achieve optimal performance across all platforms – this also applies to deep learning technologies.
“Deep Learning in machine vision promises enormous efficiency gains. On the one hand, existing applications can be improved, and on the other, entirely new applications can be automated. The collaboration with Qualcomm Technologies is a step toward providing customers with higher performance for their applications, thus securing competitive advantages through machine vision,” explains Moie. The promotion of interoperability and compatibility among components of industrial image processing is the declared goal of the MVTec Technology Partner Program. The programme is a collaborative initiative to foster strategic collaborations with innovative technology providers of various components.
