Tensorflow Image classification get train_images/train_X and train_labels/train_y

Issue

I’m working on a tensorflow model for identifying different butterfies. I’m using Neural networks for this and I’m reading images from folders and all the data gets split in a train dataset and validation dataset, but I want split these like this:

(train_images, train_labels), (test_images, test_labels) = datasets.cifar10.load_data()

instead of:

train_ds = utils.image_dataset_from_directory(data_dir, validation_split=0.2, subset="training", seed=123, image_size=(img_height, img_width), batch_size=BATCH_SIZE)
val_ds = utils.image_dataset_from_directory(data_dir, validation_split=0.2, subset="validation", seed=123, image_size=(img_height, img_width), batch_size=BATCH_SIZE)

I have tried doing this, but it makes the accuracy of my model real crap so I don’t think it’s correct:

train_images = np.concatenate([x for x, y in train_ds], axis=0)
train_labels = np.concatenate([y for x, y in train_ds], axis=0)
test_images = np.concatenate([x for x, y in val_ds], axis=0)
test_labels = np.concatenate([y for x, y in val_ds], axis=0)

I have tried a lot of methods from stackoverflow, but they also don’t work.

My model:

model = tf.keras.Sequential([
   # Please reread this link for a better understanding of the data being entered:
   #https://www.codespeedy.com/determine-input-shape-in-keras-tensorflow/
   layers.Conv2D(32, (3, 3), activation='relu', input_shape=(180, 180, 3)),
   layers.MaxPooling2D((2, 2)),
   layers.Conv2D(64, (3, 3), activation='relu'),
   layers.MaxPooling2D((2, 2), strides=2),
   layers.Flatten(),
   layers.Dropout(0.2, input_shape=(180, 180, 3)),
   layers.Dense(64, activation='relu'), 
   layers.Dense(5, activation='softmax') # there are 5 classes_names/folders or 5 kinds of butterflies
])

Solution

Fixed the problem:

train_images = np.array([]).reshape((0,180,180,3))
train_labels = np.array([]).reshape(0,)
for x, y in train_ds:
  train_images = np.concatenate((train_images, x), axis=0)
  train_labels = np.concatenate((train_labels, y), axis=0)

reshape to the input shape of the images, 0 as size is necessary otherwise, you cannot add x or y.

Answered By – IceFox_Programming

Answer Checked By – Mildred Charles (AngularFixing Admin)