Gee analysis and deep learning classification code The code provides traditional machine learning in two platforms, random forest and object-oriented methods.
For the deep learning method, the code provided is in deep learning Txt. Note: part of the code of the study area shall be changed for different study areas when using the clock; The input layer characteristics of DNN and RF+DNN structures are different. It only needs to be modified at the input band.
For the research on model transfer learning, we should pay attention to the need to save the model first ------ model Save() and then call another region to reconstruct ----keras Models Load_ Model(). Finally, the model can be used to predict this area.