This session gives an overview of the challenges faced when architecting a combined driver-assist and surround-view system, where the requirements of both machine vision and human-viewed display need to be met with a common set of cameras. The discussion will cover how the sometimes contradictory requirements on data format, system bandwidth and image quality can be met with a heterogeneous compute solution-based approach, taking advantage of the capabilities and efficiencies of the individual processing elements from image sensor through ISP, GPU and CPU to display output.