Millimeter wave (mmWave) and terahertz (THz) drones have the potential to enable several futuristic applications such as coverage extension, enhanced security monitoring, and disaster management. However, these drones need to deploy large antenna arrays and use narrow directive beams to maintain a sufficient link budget. The large beam training overhead associated with these arrays makes adjusting these narrow beams challenging for highly-mobile drones. To address these challenges, this paper proposes a vision-aided machine learning-based approach that leverages visual data collected from cameras installed on the drones to enable fast and accurate beam prediction. Further, to facilitate the evaluation of the proposed solution, we build a synthetic drone communication dataset consisting of co-existing wireless and visual data. The proposed vision-aided solution achieves a top-1 beam prediction accuracy of ≈91% and close to 100% top-3 accuracy. These results highlight the efficacy of the proposed solution towards enabling highly mobile mmWave/THz drone communication.