[FIXED] Display an image and its label from DataFrameIterator

Issue

I am iterating images from a dataset(created with flow_from_dataframe) and would like to display the image using pyplot and its corresponding label

training_dataset = train_dataset_gen.flow_from_dataframe(dataframe=train_df,
                                                     directory="./Data/train/",
                                                     x_col="file",
                                                     y_col="label",
                                                     shuffle=True,
                                                     batch_size=BATCH_SIZE,
                                                     target_size=IMG_SIZE,
                                                     class_mode="sparse",
                                                     subset="training")

I am able to display the image but the label is an array of 0's and 1's.

for _ in range(len(training_dataset.filenames)):
     image, label = training_dataset.next()
     # display the image from the iterator
     plt.imshow(image[0])
     plt.title(training_dataset.)
     plt.show()

How do I get the real label value?

This was solved by reversing the dictionary:

class_names=training_dataset.class_indices

new_dict={}
for key, value in class_names.items():
    new_dict[value]=key

Solution

ImageDataGenerator has an attribute .class_indices. It is a dictionary of the form

{'class name': class index}

where class name is a string which is the name of the label in your dataframe label column. Class index is an integer associate with the class name. This is the values you will see when you print the labels (since you used class_mode='sparse'). For example in your dataframe if you labels were cat, dog then your class_indices would be

{'cat':0, 'dog':1}

It is convenient to reverse this dictionary as shown below

for key, value in class_dict.items():
    new_dict[value]=key 

now new_dict is of the form {0:'cat', 1:'dog'}. So now to modify your code as follows

for _ in range(len(training_dataset.filenames)):
     image, label = training_dataset.next()
     # display the image from the iterator
     plt.imshow(image[0])
     label_name=new_dict[label[0]] # note you are only showing the first image of the batch
     plt.title(label_name)
     plt.show()

Answered By - Gerry P