Small sample SAR (Synthetic Aperture Radar) target identification method based on two-stage contrast learning framework

The invention discloses a small sample SAR target recognition method based on a two-stage contrast learning framework, and relates to image recognition. In order to solve the practical problems that an SAR data set is difficult to obtain and time and labor are consumed in data labeling, a contrast learning framework based on a two-stage training strategy is provided and is used for solving SAR target recognition under the condition of small sample data. A training process is decoupled into a representation learning stage and a classification learning stage, and the method comprises the following steps: 1) in the representation learning stage, training a representation module of a network on an original SAR data set by adopting supervised contrast learning; and 2) a classification learning stage: fixing a coding module and training a classification module of the network by adopting a heavy balance strategy. The method is based on comparison learning in machine learning, combines the idea of decoupling learning