Existing research in the field of smart farms has only been studied to increase production by maintaining the optimal growing environment for crops. However, in the case of tomatoes, the harvest time considering the shelf life is very important, so quantified harvest time prediction information is needed. Therefore, this paper designed a system to predict the exact harvest time by measuring the temperature of the fruit directly with a thermal imaging camera rather than the accumulated temperature through the existing greenhouse temperature. A database was built by collecting thermal image data and environmental data, and a system was designed to predict the harvest time of crops through a linear regression algorithm and integrated temperature calculation formula. Through this system, it is possible to predict the exact harvest time of crops, and furthermore, it is possible to control the production time by controlling the smart farm environment, which has the advantage of shipping crops at a high selling price. Therefore, farmers can see the effect of increasing profits through high-quality crops.