Smart cameras demand high image processing capability, low power consumption, and functional safety.
In order to achieve this, the R-Car V3M uses a Renesas original image processing engine (IMP) and a new computer vision engine (CVE) that supports deep learning. This gives it the powerful computational processing capabilities needed to recognize product/road signs, lanes, and other vehicles in real-time. In terms of functional safety, it conforms to ISO26262 ASIL-C.
This video demonstrates the difference in power consumption between CPU-only processing and IMP-based processing.