Chemical fertilizer particle instance segmentation method based on improved Mask R-CNN

The invention discloses a chemical fertilizer particle instance segmentation method based on an improved Mask R-CNN. The method mainly comprises the following four steps: collecting chemical fertilizer particle images and constructing an image library; performing enhancement, preprocessing and data set labeling on the image, and then dividing a training set and a test set; improving a Mask R-CNN network and training a chemical fertilizer particle segmentation model based on the network; and performing instance segmentation on the chemical fertilizer particle image in the test set based on the trained segmentation model, and obtaining an instance segmentation result. Compared with a traditional segmentation network, the improved Mask R-CNN network greatly improves the accuracy of chemical fertilizer particle segmentation. According to the method provided by the invention, the defects and deficiencies of chemical fertilizer particle segmentation in the prior art can be overcome, and a basis is provided for part