The repository gives an example of SAR ATR on public MSTAR dataset with machine learning, using python 3.6 and scikit-learn API. It is recommended to directly install Anaconda 3 which provides common package like numpy, matplotlib and scipy as well as scikit-learn.
The example supports implementation with most machine learning methods including logistic regression, neural network, support vector machine and so on. For reducing computation, it uses principle components analysis for feature reduction.
Data processing and model declaration are respectively defined in data.py and model.py. Of cource you can add your own manipulation.