Growing demand for automotive cameras necessitates intelligent vision processors.

CEVA announced that it will provide its imaging and vision platform to ON Semiconductor’s automotive Advanced Driver Assistance System (ADAS) product family.

ON Semiconductor said it will leverage the new image processing capabilities of CEVA's vision processing IP to enhance the performance of its future products and integrate embedded intelligence and machine learning into its ADAS roadmap.

Strategy Analytics predicts that demand for automotive cameras will surpass 200 million units by 2023, driven by the emergence of new camera applications and imaging concepts, and that the role of intelligent vision processors in automotive safety systems will evolve rapidly.

Additionally, intelligent vision processors are cost-effective because they come with advanced safety features available at various price points, making them suitable for widespread adoption in the automotive market.

Efficient vision processing can address computationally intensive imaging and machine learning challenges, enabling improved low-light image processing and the execution of deep neural networks. For these reasons, ON Semiconductor selected CEVA's imaging and vision platform to bolster its ADAS products and capitalize on this rapidly growing market opportunity.

Efficiently address video analytics, augmented reality, ADAS, and deep learning application requirements.

CEVA's imaging and vision platforms efficiently address the intensive processing requirements of sophisticated computer vision and deep learning applications such as video analytics, augmented reality, and ADAS. These high-efficiency vision DSPs (Digital Signal Processors) and accelerators reduce the power consumption of the entire system.

CEVA's XM DSPs incorporate a hybrid architecture consisting of vector DSP processors and scalars, combined with a comprehensive Application Development Kit (ADK) to simplify software deployment.

CEVA's ADK includes ▲CEVA-Link for software-level integration with the host processor ▲a range of widely used and optimized software algorithms ▲the second-generation deep neural network software framework (CEVA Deep Neural Network, CDNN2) that simplifies the introduction of machine learning at a fraction of the power consumption of leading GPU-based systems ▲state-of-the-art development and debugging tools.