Fecha de publicación: 12/05/2023 Fuente: WIPO "processing tomato" The invention discloses a CBAM attention mechanism tomato detection method based on YOLOv7. The CBAM attention mechanism tomato detection method has the advantages of reducing noise, improving recognition accuracy and the like. Firstly, labeling, data enhancement and data set division are carried out on an acquired greenhouse string tomato image, secondly, a Gaussian filter is introduced before the image is input into a target detection backbone network for feature extraction, noise reduction processing is carried out on the image, then a CBAM module is introduced into a YOLOv7 model after feature extraction of the backbone network is finished, and a CBAM model is introduced into the YOLOv7 model. And feature enhancement is sequentially carried out from a channel domain and a spatial domain, so that the identification accuracy of the string-type tomatoes is further improved. According to data display, the recognition accuracy can reach 95.8%, the field recognition rate can reach 91%, the method meets practical requirements in the aspects of detection accuracy and detection speed, and development of tomato picking robots and facility agriculture is further promoted.