How to define a Pandas column as a list-CodePudding

I'm using Keras Image Data Generator for data augmentation, and the flow_from_dataframe function within it. Info regarding it here: https://www.tensorflow.org/api_docs/python/tf/keras/preprocessing/image/ImageDataGenerator#flow_from_dataframe

# Create new dataframes for train and test

df_train = pd.DataFrame()
df_train['image'], df_train['labels'] = X_train, y_train

df_test = pd.DataFrame()
df_test['image'], df_test['labels'] = X_test, y_test

This is what one dataframe looks like:

image   labels
4227  /Users/m/Documents/Machine Learning Pr...  [73, 0]
4676  /Users/m/Documents/Machine Learning Pr...  [36, 0]
800   /Users/m/Documents/Machine Learning Pr...  [26, 0]
3671  /Users/m/Documents/Machine Learning Pr...  [42, 0]

This is how I've imported the data generator:

from keras.preprocessing.image import ImageDataGenerator

datagen = ImageDataGenerator(
    rescale = 1./255,
    rotation_range = 40,
    width_shift_range = 0.2,
    height_shift_range = 0.2,
    shear_range = 0.2,
    zoom_range = 0.2,
    horizontal_flip = True,
    fill_mode = 'nearest'
    )

test_datagen= ImageDataGenerator(rescale=1./255.)

train_generator=datagen.flow_from_dataframe(
dataframe = df_train,
x_col="image",
y_col="labels",
batch_size=32,
seed=42,
shuffle=True,
class_mode='multi_output',
target_size=(128, 128))


valid_generator = test_datagen.flow_from_dataframe(
dataframe = df_test,
x_col = "image",
y_col = "labels",
batch_size = 32,
seed = 42,
shuffle = True,
class_mode='multi_output',

target_size=(128, 128))

The function reads in a dataframe, but in the documentation it says the y_col specified must be a list:

y_col string or list, column/s in dataframe that has the target data.

Before I created the dataframe the column was a list, but now that it's a column in pandas it's no longer classed as a 'list', right? So why do I get this error message:

TypeError: If class_mode="multi_output", y_col must be a list. Received str.

I want to use the class mode multi outputas above, and it states y_col must be a list but it's a string. Not sure why it is stating it's a string? Is there anyway to change the 'type' of a column within a dataframe or am I misunderstanding?

CodePudding user response：

'List' here means list of column names.

As Zelemist has said, change your dataframe so that there are two columns rather than the one you have.

Then input a list to y_col such as:

y_col = ['col1', 'col2]

Hope it makes sense now.