lim-anggun / keras-imagedatagenerator Goto Github PK

I'm trying to make an AutoEncoder using the keras ImageDataGenerator function. Now because I want to build an Autoencoder, I dont require the image labels because the Y_train of my dataset are not labels but images itself.
When I try to fit my model and pass X_train and Y_train it shows an error :
y argument is not supported when using keras.utils.Sequence as input.

How can I get away with this.

Thanks for your post. It's very helpful for me.
But during running the code for CamVid dataset, I got an error syaing "AttributeError: 'tuple' object has no attribute 'ndim'". I think this comes from the "zip" function. But I have no idea about a solution.

Did you get the same issue as mine?

conv2d
C:\Users\Chinn\AppData\Local\Temp\ipykernel_22868\1831489917.py:14: RuntimeWarning: invalid value encountered in true_divide
channel_image /= channel_image.std()
C:\Users\Chinn\AppData\Local\Temp\ipykernel_22868\1831489917.py:29: UserWarning: Attempting to set identical bottom == top == -0.5 results in singular transformations; automatically expanding.
plt.imshow(display_grid, aspect='auto', cmap='viridis')
average_pooling2d
conv2d_1
average_pooling2d_1
conv2d_2
average_pooling2d_2
conv2d_3
average_pooling2d_3
flatten

MemoryError Traceback (most recent call last)
Input In [16], in <cell line: 1>()
4 size = layer_activation.shape[1] #The feature map has shape (1, size, size, n_features).
5 n_cols = n_features // images_per_row # Tiles the activation channels in this matrix
----> 6 display_grid = np.zeros((size * n_cols, images_per_row * size))
7 for col in range(n_cols): # Tiles each filter into a big horizontal grid
8 for row in range(images_per_row):

MemoryError: Unable to allocate 64.0 GiB for an array with shape (262144, 32768) and data type float64

Hi,
Do you know how to implement flow from directory when using .npy files?

Best,
Kirsten