The conventional approach to the implementation of embedded vision systems relies on “black boxes” developed upon prescribed specifications. This can lead to suboptimal designs and, in the worst case, to unrealizable requirements for some of the system components. ENVISAGE is a research project that will follow a holistic approach with the aim of boosting performance and increasing power efficiency with respect to the conventional approach. This alternative path demands a deep knowledge of the impact of design parameters at various levels of the system hierarchy. In fact, multi-level optimization is mandatory to meet high-level specifications efficiently. The broad scope and interdisciplinary view required to address multi-level design and optimization distinctly call for a critical mass of researchers with expertise in different areas. Still, they must share a common baseline knowledge to properly convey and understand specifications over an extensive range of abstraction layers. Research teams at IMSE, CiTIUS-USC, and UPCT have merged their skills and background to cover this global perspective and accomplish the ambitious objectives of ENVISAGE (RTI2018-097088-B).