Yolo v8-DPE-based potato impurity identification and positioning method

The invention relates to the technical field of potato harvesting, in particular to a potato impurity identification and positioning method based on Yolo v8-DPE, and the method comprises the steps: S1, collecting an image in real time in an impurity removal process; s2, pre-processing the image to reduce image noise, wherein the pre-processing comprises cutting and filtering; s3, performing instance segmentation on the potato impurities on the preprocessed image by using a Yolo v8-DPE algorithm; s4, outputting a mask layer picture after instance segmentation, wherein instances and backgrounds are marked by adopting different colors; s5, noise reduction processing is carried out on the mask layer picture, and targets with the pixel area smaller than a preset value in the picture are removed; s6, determining the center of gravity of the target potato impurities; s7, determining a clamping point of the target potato impurities; and S8, extracting depth information of the target potato impurity clamping point, an