Method and system for automatically marking, classifying and counting pollen image features

Abstract

The invention relates to the technical field of artificial intelligence and image processing, in particular to a pollen image feature automatic labeling, classifying and counting method and system, and the method comprises the steps: obtaining a time sequence microscopic image sequence, executing the instance segmentation of an initial time frame of the time sequence microscopic image sequence, and obtaining an initial time frame of the time sequence microscopic image sequence; identifying each independent pollen individual and distributing a globally unique identity identifier for each independent pollen individual; based on the identity identifier, using a non-rigid motion tracking technology to associate the segmented instances of the same pollen individual in the continuous time frames of the time sequence microscopic image sequence, and generating a development trajectory of each pollen individual; extracting a germination event time point of each pollen individual and a phenotypic feature vector representing development dynamics from the development trajectory of each pollen individual, and counting the number of germinated pollen and the number of non-germinated pollen according to the germination event time point; and inputting the phenotypic feature vector into a pre-trained classification model, and outputting a pollen viability category label.

Images