Keras：如何使用`image_dataset_from_directory`加载测试集？

Question

I am using tf.keras.preprocessing.image_dataset_from_directory to load dataset as follows,

train_dataset = tf.keras.preprocessing.image_dataset_from_directory(train_dir, 
                                                                    labels='inferred', 
                                                                    label_mode='categorical',
                                                                    batch_size=32,
                                                                    image_size=(224, 224))


val_dataset = tf.keras.preprocessing.image_dataset_from_directory(val_dir, 
                                                                  labels='inferred', 
                                                                  label_mode='categorical',
                                                                  batch_size=32,
                                                                  image_size=(224, 224))

However, when I check the document looks like this argument labels seem to be a must-have one, but my test data has no labels, so how can I load test data? Is there a convenient and unified way to do this?

Answer 1

If your data isn't labeled, I don't think you can call it the test set , since you won't be able to evaluate the performance of your algorithm using it.

The argument you're looking for is label_mode , see the documentation . If you set it to label_model=None , it will not return a target;

label_mode : 'int' : means that the labels are encoded as integers (eg for sparse_categorical_crossentropy loss). 'categorical' means that the labels are encoded as a categorical vector (eg for categorical_crossentropy loss). 'binary' means that the labels (there can be only 2) are encoded as float32 scalars with values 0 or 1 (eg for binary_crossentropy). None (no labels).