Smart agriculture is a trendy topic as it has a clear impact in both productivity, ecological impact, and improvement of working conditions. Smart viticulture is one of the domains that can benefit both from wireless sensor networks and mobile devices embarked in vineyard labor tools (e.g., on a straddler tractor). One important use case is related to the yield estimation, an invaluable information to drive the harvest organization, plant management, and business’s economy. Traditional methods rely on destructive sampling and manual counting, resulting in error rates sometimes greater than 30%. In this chapter, we review existing techniques for the automation of yield estimation and, focusing on deep learning methods, propose some strategies and preliminary results obtained in a production environment.