Abstract
An innovative and flexible system for performing terrestrial meteorological and solar measurements based on computer vision technology and edge artificial intelligence is presented here. The proposed reference architecture hardware was developed in the context of the DeepSky project and is able to deliver a fully working system for acquiring all sky images from different vision modalities and monitoring the meteorological and solar variabilities through the inspection of cloud instances in real-time. The corresponding software is able to provide cloud classification feedback, coverage percentage and depict weather statistics though a novel and efficient user interface.