One of the most pressing problems in agricultural science is understanding global stored seed bank variation and how it might be successfully deployed to feed a growing population. Less than 1% of accessions held by Plant Genetic Resources for Food and Agriculture are utilized for crop improvement. Phenomics, a science that integrates genotype, environment, and time, has the potential to characterize stored genetic variation in genebanks. Phenomic approaches that make use of digital imaging and machine learning may provide high-quality data that can be used for germplasm assessment. However, much of these data at present are neither centralized nor attached to germplasm repositories. An enormous opportunity exists to integrate physical collections of seeds with phenomic information from those accessions. A global approach to centralizing phenomic data on crop germplasm may require the cooperation of world governments and require substantial resources, but the potential to more fully characterize our crop genetic resources using these techniques should foster improved opportunities for plant breeders to increase food production.