Viruses represent the most numerous ‘‘biological entity,’ on Earth, that are the unseen driver of global biogeochemical processes, through viral lysis, selection, and genetic exchange. Globally viral abundance has been estimated to be the 10³¹, known as the ‘Hendrix product’ which predicts that viral abundance (via viral-like particles - VLPs) is larger then the stars in the observable universe (10²¹). Counting viruses in aquatic, soils, sediments, vertebrates, and mats have had ongoing methodological issues. Such issues include time consuming human counting by eye to enumerate viral abundance, and also lacks accurate sizing to eliminate false positives/negatives, generated by human error. To obtain better estimates of VLPs, we have developed a novel intelligent algorithm for computer vision via a convolutional neural network (CNN). This automated software CYCLOPS resolves VLPs within microscopy images using a cost-effective, user-friendly method for pennies on the dollar. Our software with modified microscopy methods provides viral abundance, viral size, and nucleic acid content abundance. We have furthermore streamlined the sample processing to better estimate viral abundances across multiple biomes including soils/rhizosphere, roots, mats, and vertebrates (e.g., fish). We can accurately size viruses from 50 - 500 nm at high resolution within a complex environmental sample in <1 minute per image. Our method also enumerates giant viruses >500 nm with a novel sample preparation that removes bacteria but keeps large viruses intact for counting. Still, challenges remain to complete enumeration for VLPs which include: 1) specific dyes for ssDNA and RNA within a dsDNA VLP mixture, and 2) fixation and preparation for membranes/lipid-containing VLPs. Our computer vision software combined with single virion sequencing with our could illuminate the unseen viral diversity present on our planet.