Keras ImageDataGenerator Low Validation Accuracy - keras

Question

I'm trying to preform transfer learning with mobilenetv2 to classify 196 classes of cars from the cars-196 dataset of stanford.
My work environment is google colab notebook.
I use the ImageDataGenerator from keras to load the images for the train and validation.
On the training images I also perform data augmentation.
The following code is how I perform it:
# To load the dataset from the drive
from google.colab import drive
drive.mount('/content/drive')
import math
from keras.models import Sequential, Model
from keras.layers import Dense, Conv2D, MaxPooling2D, Flatten, Dropout, ReLU, GlobalAveragePooling2D
from keras.preprocessing.image import ImageDataGenerator
from keras.applications.mobilenet_v2 import MobileNetV2, preprocess_input
BATCH_SIZE = 196
train_datagen = ImageDataGenerator(
rotation_range=20, # Rotate the augmented image by 20 degrees
zoom_range=0.2, # Zoom by 20% more or less
horizontal_flip=True, # Allow for horizontal flips of augmented images
brightness_range=[0.8, 1.2], # Lighter and darker images by 20%
width_shift_range=0.1,
height_shift_range=0.1,
preprocessing_function=preprocess_input
)
img_data_iterator = train_datagen.flow_from_directory(
# Where to take the data from, the classes are the sub folder names
'/content/drive/My Drive/Datasets/cars-196/car_data/train',
class_mode="categorical", # classes are in 2D one hot encoded way, default is true but just to point it out
shuffle=True, # shuffle the data, default is true but just to point it out
batch_size=BATCH_SIZE,
target_size=(224, 224) # This size is the default of mobilenet NN
)
validation_img_data_iterator = ImageDataGenerator().flow_from_directory(
'/content/drive/My Drive/Datasets/cars-196/car_data/test',
class_mode="categorical",
shuffle=True,
batch_size=BATCH_SIZE,
target_size=(224, 224)
)
base_model = MobileNetV2(weights='imagenet', include_top=False)
x = base_model.output
x = GlobalAveragePooling2D()(x)
x = Dense(512, activation='relu')(x)
x = Dropout(0.5)(x)
preds = Dense(196, activation='softmax')(x)
model = Model(inputs=base_model.input, outputs=preds)
# Disable training of already trained layer
for layer in model.layers[:-3]:
layer.trainable = False
model.compile(optimizer='Adam',loss='categorical_crossentropy',metrics=['accuracy'])
# define the checkpoint
from keras.callbacks import ModelCheckpoint
filepath = "/content/drive/My Drive/Datasets/cars-196/model.h5"
checkpoint = ModelCheckpoint(filepath, monitor='loss', verbose=1, save_best_only=True, mode='min')
callbacks_list = [checkpoint]
history = model.fit(
img_data_iterator,
steps_per_epoch=math.ceil(8144/BATCH_SIZE), # 8144 is the number of training images
validation_steps=math.ceil(8062/BATCH_SIZE), # 8062 is the number of validation images
validation_data=validation_img_data_iterator,
epochs=100,
callbacks=callbacks_list
)
About the batch size, from this stackoverflow question I decided to set the batch size as the number of labels available, but it did not change anything in terms of val_accuracy.
I've added a dropout of 0.5 between the fully connected layers that I have added, but again, no change in the accuracy of the validation.
My accuracy on the training set arrives to about 92% while the validation accuracy stays at about 0.7%.
My guess is that the ImageDataGenerator is acting weird and screwing up the accuracy, but I have not found any solution for the problem so ATM I dont have a clew what is the reason behind it.
Does anyone have any guesses as to what might be the problem?
----- EDIT
The train and test folder all have sub folders with name of the labels (the different cars I want to identify) and each subfolder has images of that car. This is just how the cars-196 dataset is. The ImageDataGenerator attaches the right label to the image, depending on in what subfolder that image was in.

Answer 1

The problem was that I did not apply the preprocess_input function to the validation data image generator.
Instead of
validation_img_data_iterator = ImageDataGenerator().flow_from_directory(
'/content/drive/My Drive/Datasets/cars-196/car_data/test',
class_mode="categorical",
shuffle=True,
batch_size=BATCH_SIZE,
target_size=(224, 224)
)
Changed it to
validation_img_data_iterator = ImageDataGenerator(
preprocessing_function=preprocess_input
).flow_from_directory(
'/content/drive/My Drive/Datasets/cars-196/car_data/test',
class_mode="categorical",
shuffle=True,
batch_size=BATCH_SIZE,
target_size=(224, 224)
)