Using CEVA’s CDNN software framework, DeepGlint achieves 5x performance/power ratio advantage and affordable embedded solution versus GPU-based system, at a fraction of the cost and power consumption; Technology on display at CES 2017
CEVA, Inc. (NASDAQ: CEVA), the leading licensor of signal processing IP for smarter, connected devices, and DeepGlint Technology, Ltd., a world leader in the detection, tracking and recognition of humans and vehicles, announced today that DeepGlint has made available its industry-leading security surveillance and intelligent traffic algorithms on the CEVA imaging and vision platform.
DeepGlint utilized the CEVA Deep Neural Network (CDNN) software framework to generate an implementation of their neural networks for the CEVA-XM4 DSP, that deliver outstanding performance and power efficiency for deep learning applications. The solution is ideal for low-cost embedded vision systems where GPU-based systems are not feasible.
As we continuously seek to improve our product value proposition, we required a vision processor architecture that can handle our complex neural networks within stringent cost and power budgets. The CEVA-XM4 platform which incorporates the CDNN software framework excels in this regard and opens up a much larger market opportunity with our computer vision and deep learning technologies in the fast growing embedded vision space.
Yafeng Deng, CTO of DeepGlint,
Ilan Yona, vice president and general manager of the Vision Business Unit at CEVA, commented: “DeepGlint is a highly respected leader in computer vision and machine learning and has managed to harness the full potential of our imaging and vision platform to prove undoubtedly and independently that our deep learning platform not only stands toe to toe with GPU-based systems, but offer many clear advantages over them. We look forward to expanding our partnership with them to address the needs of our mutual customers.”
CEVA will showcase the DeepGlint technology at the upcoming Consumer Electronics Show (CES) 2017, January 5-8, 2017 in Las Vegas. To schedule a meeting, contact CEVA at [email protected].