Fecha de publicación: 01/09/2023 Fuente: WIPO "processing tomato" The invention provides an inspection robot real-time position identification and tomato counting yield evaluation method, which comprises the following steps: a ridge row coordinate information acquisition step: an inspection robot moves circularly among a plurality of tracks to acquire ridge row coordinate information; tomato inspection and counting: the inspection robot acquires tomato images in real time, tomato targets with different maturity are obtained through deep learning algorithm processing, target tracking is realized based on feature matching, and the number of tomatoes is counted; and a position number association step: establishing a relationship between the counting number of mature tomatoes on the current ridge and the current track position. Precise orbit determination is achieved through multiple sensors, position coordinates are accurate, and consumed time is short. According to the method, tomato targets are identified through a deep learning algorithm, and target tracking is realized through feature matching, so that a multi-target tomato counting task of different maturity is completed, and repeated counting of tomatoes is avoided. According to the method, the track position and the tomato number are fused, so that the tomato number of the inspection robot under each coordinate at each moment is obtained.