Convolutional neural network model design space McaNetX and optimization method thereof

The invention provides a convolutional neural network model design space McaNetX and an optimization method thereof. The optimization method comprises the following steps: 1) designing a basic structure of a feature extraction network according to characteristics of an SAR image, and representing the basic structure by using structure parameters; 2) analyzing the structure parameter rule in the step 1) by utilizing a design space sample analysis tool; 3) integrally evaluating the performance of the design space sample model, and comparing the change of the performance; and 4) optimizing the design space set McaNetX through the structure parameter change rule in the step 2). The network structure designed based on the design space optimization method can complete the target identification task of the SAR image on the lightweight model, and has strong robustness and generalization ability. 本发明提供了一种卷积神经网络模型设计空间McaNetX及其优化方法，所述优化方法包括以下步骤：1)根据SAR图像的特点设计特征提取网络的基础结构并将基础结构使用结构参数表示；2)利用设计空间样本分析工具分析步骤1)中的结构参数规律；3)整体评判设