This invention discloses a method and
system for posture
estimation and
cutting point localization of
rhizome medicinal materials, belonging to the field of
digital agriculture technology. The method includes: acquiring training data; constructing a YOLO model; training recognition and localization models for each part; inputting the
RGB image of the
rhizome medicinal material to be identified into the model; preprocessing the image using a multi-frame optimization strategy, obtaining the contour of the largest connected region after denoising, and calculating the
centroid; using a geometric constraint strategy, recovering the root-stem transition zone detection box based on the spatial position of the root and stem, obtaining its largest connected region contour, and calculating the
centroid; based on the root-stem transition zone and the optimized stem and root detection boxes and centroids, using a root-stem opposite-direction clipping
algorithm and weighted
least squares fitting iteration to obtain the posture angle of the
rhizome medicinal material; finally, locating the
cutting point of the rhizome medicinal material based on the posture angle and the structural geometric features of the root-stem transition zone. This invention offers the advantages of accurate, efficient, and rapid posture
estimation and
cutting point localization of rhizome medicinal materials, enabling automated root-stem separation and
processing of rhizome medicinal materials.