I need help to use only 8 out of 10 classes from cifar 10. Normally it loads all 10 classes

Question

I want to perform 8 class classification only and hence need to filter any 8 classes out of 10. Please help. Thank you!

Code to load cifar 10 is below

#Keras library for CIFAR-10 dataset
from keras.datasets import cifar10


#Downloading the CIFAR dataset
(x_train,y_train),(x_test,y_test)=cifar10.load_data()

# I tried the following but its changing the array shape please help
#Train
for i in range(8):
  index = np.where(y_train == i)
  X_train = x_train[index]
  Y_train = y_train[index]

#Test
for i in range(8):
  index = np.where(y_test == i)
  X_test = x_test[index]
  Y_test = y_test[index]

Answer 1

(X_train, y_train), (X_test, y_test) = tf.keras.datasets.cifar10.load_data()

this contains all samples of 10 classes

you can choose index respect to classes

index = np.where(y_train.reshape(-1) == 0)
X_train = X_train[index]
y_train = y_train[index]

This gives all the samples with a label of 0

You can use this technique to load any specific classes you want.

Or if you want to load multiple classes, you can specify multiple conditions, like this:

class_0_index = np.where(y_train.reshape(-1) == 0)
X_train_class_0 = X_train[class_0_index]
y_train_class_0 = y_train[class_0_index]

class_1_index = np.where(y_train.reshape(-1) == 1)
X_train_class_1 = X_train[class_1_index]
y_train_class_1 = y_train[class_1_index]

class_2_index = np.where(y_train.reshape(-1) == 2)
X_train_class_2 = X_train[class_2_index]
y_train_class_2 = y_train[class_2_index]

X_train = np.concatenate((X_train_class_0, X_train_class_1, X_train_class_2))
y_train = np.concatenate((y_train_class_0, y_train_class_1, y_train_class_2)).reshape(-1,1)

This code loads all the samples with a label of 0, 1 and 2