CODEWITHSUNDEEP
pythontensorflowmachine-learningkerasconv-neural-network
Mirza Munib Baig
Mirza Munib Baig

Reputation: 311

Confused between whether or not to change the class_mode from binary to categorical in Imagedatagenerator

When i shifted from 2 classes to 4 classes in a CNN model. I got errors. A lot of people helped me solve them, like I needed to change loss to sparse_categorical_crossentropy and last layer activation function.

model.compile(loss='sparse_categorical_crossentropy',
              optimizer='adam',
              metrics=['accuracy'])

history = model.fit_generator(
    train_generator,
    steps_per_epoch=nb_train_samples//batch_size,
    epochs=epochs,
    validation_data = validation_generator, 
    validation_steps = validation_generator.samples // batch_size,
)

I am confused as I had to change the loss function from binary_crossentropy to categorical_crossenpropy. I want to know whether I should change the class_mode from binary to categorical for 4 classes or should I leave it as binary.

train_datagen=ImageDataGenerator(
    rescale=1./255,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip = True,
    #vertical_flip = True,
    validation_split=0.2,
    brightness_range=[0.5, 1.5]

)

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

train_generator = train_datagen.flow_from_directory(
    train_data_dir,
    target_size=(img_width,img_height),
    batch_size=batch_size,
    shuffle=True,
    class_mode='binary',
    subset='training'
)

validation_generator = train_datagen.flow_from_directory(
    train_data_dir, # same directory as training data
    target_size=(img_height, img_width),
    batch_size=batch_size,
    class_mode='binary',
    subset='validation'
    #validation_data_dir,
    #target_size=(img_width,img_height),
    #batch_size=batch_size,
    #class_mode='binary'
)

Upvotes: 2

Views: 3851

Answers (2)

Nicolas Gervais
Nicolas Gervais

Reputation: 36604

I think reading the documentation would be a good starting point. It would answer all your questions, and then some.

class_mode: One of "categorical", "binary", "sparse", "input", or None. Default: "categorical". Determines the type of label arrays that are returned: - "categorical" will be 2D one-hot encoded labels, - "binary" will be 1D binary labels, "sparse" will be 1D integer labels, - "input" will be images identical to input images (mainly used to work with autoencoders). - If None, no labels are returned (the generator will only yield batches of image data, which is useful to use with model.predict_generator()). Please note that in case of class_mode None, the data still needs to reside in a subdirectory of directory for it to work correctly.

So you should use categorical_crossentropy as loss function if you choose categorical for class_mode, and sparse_categorical_crossentropy if you choose sparse.

Upvotes: 3

Yoskutik
Yoskutik

Reputation: 2079

Of course, you should change it. If you have only 2 classes use class_mode='binary' in flow_from_directory function and binary_crossentropy as loss. In case you have more than 2 classes, use class_mode='categorical' and categorical_crossentropy

Upvotes: 2

Related Questions