从我自己的图像数据中获得xtrain，y train，xtest和ytrain

Question

My data is ordered in folders as follows:

- train
    * principal
    * soup
    * dessert
- test
    * principal
    * soup
    * dessert

It contains 1750 64x64 img per folder in train and 250 p/folder in test. I want to obtain xtest / ytest / xtrain / ytrain vars from that data. But I only find a way to do it from, for example CIFAR-10:

(x_train, y_train), (x_test, y_test) = cifar10.load_data() 
# x_train - training data(images), y_train - labels(digits)

Any thought? Thanks.