Prediction of Harvest Time of Tomato Using Mask R-CNN

In recent years, the agricultural field has been confronting difficulties such as aging of farmers, a shortage workers, and for new farmers. Harvesting time prediction potential to solve these problems, is expected effectively utilize human resources, save labor, reduce labor costs. To achieve harvesting prediction, various works are being actively conducted. Methods using meteorological information temperature solar radiation, etc., methods neural networks based on color from fruit bunch images investigated. However, accuracy still insufficient, individual tomato fruits not studied. this study, we propose novel method predict fruits. The uses Mask R-CNN detect bunches two types ripeness determination experimental results showed that ratio R values was better tomatoes close time, average differences between G in RGB far time. These show effectiveness proposed method.