Orchard sweet cherry color distribution estimation from wireless sensor networks and video-based fruit detection

Luis Cossio-Montefinale, Cristóbal Quiñinao, Rodrigo Verschae

Abstract: Highlights•A Novel methodology is proposed that combines agroclimatic wireless sensor network data with video-based fruit detection and tracking for assessing cherry development.•Computer vision techniques based on deep learning and Kalman Filters (for fruit detection and tracking) are used for fruit counting and system calibration•Climate data transmitted through a LoRaWAN network and modeling spatial dynamics with a KNN is used to estimate fruit color distribution•Two methods for assessing maturity distribution are presented, with the second method, relying solely on sensor network data, achieving an 5% MSE error rate.•The proposed methods are validated using data from productive fields.