Kernel markers (i.e., easily scorable phenotypes) have been used to study inheritance in maize for at least a hundred years, but methods to track these markers have changed little over time. Here we present a novel method to phenotype maize ears using a custom-built rotational scanner. Our cost-effective scanning system creates a 2D projection of the ear, which can then be processed using an internally-developed computer vision and machine learning pipeline to identify kernel locations and corresponding markers. This positional data can be further examined to assess spatial patterns of marker inheritance across the maize ear, providing an improved method to identify mutants that alter such patterns. As a test case, the transmission rates of dozens of GFP-tagged mutations potentially affecting the maize male gametophyte (pollen) were quantified, identifying several genes with putative male-specific functions in reproduction. This methodology enables phenotypes to be analyzed on the scale of thousands of kernels across multiple ears, providing sufficient data to undertake robust statistical analyses and identify subtle but significant deviations from Mendelian inheritance associated with mutations in specific genes.