In this work, we consider Reconfigurable Intelligent Meta-surfaces (RIMs) to provide effective beam steering functionality, while enhancing the coverage of the fifth generation (5G) users in a urban city mobility context. In particular, we enable communications in the millimeter wave (mmWave) frequencies, that can occur simultaneously with other 5G communications. The main objective of this paper is to demonstrate the feasibility and advantages of an integrated Computed Vision (CV) approach for tuning the meta-atom states of a RIM structure working at mmWave frequency band, without estimating the interferes' radio channel. In particular, the CV system will feed a logic unit running a Machine Learning (ML) algorithm to compute in real-time the coding schemes, namely the sequence of binary states associated to each unit-cell of the RIM, in order to obtain the target reconfigurable radiation pattern. Specifically, a Genetic Algorithm (GA) is introduced to derive the most suitable radiation pattern for Beam Steering application, based on the input of a CV infrastructure. Results show the feasibility of such a kind of system, with an higher coverage achieved in dense scenarios, by improving the robustness against the potential blockages introduced by the mmWave technology. Moreover, we demonstrate the system is robust against the inaccuracy introduced by the CV and GA.