Roman Solovyev

I think you need to rewrite WBF in Java to use with this framework. Algorithm is pretty simple and I think it's not a problem. Probably I should make a...

No I didn't embed it in Yolo V5. This code is universal and can be used with any object detection models.

Can you please provide tensorflow and keras version?

I changed your code like that. And it works totally fine for me: ``` def get_simple_2d_model(): import tensorflow as tf import tensorflow.keras as keras from tensorflow.keras.layers import MaxPool2D, Flatten, ReLU...

May be it's because this line? x = Dense (512, activation = tf.nn.relu,kernel_regularizer=l2(0.01))(x) There are direct TF calls.

I made support for tensorflow.keras. It must work fine now.

I made support for tensorflow.keras several weeks ago. It must work fine now. If you have problem please post error message.

@Choapinus do you have some custom layers in your model? Also check this: https://stackoverflow.com/questions/58878421/unexpected-keyword-argument-ragged-in-keras

Yes, it's batchnorm fusion.

Yes, it's probably the same, but there are not enough details in description.