How to understand NN input and Output

Question

I have a project where I want to do feature extraction in images. I have this CNN layer:

model = keras.Sequential(
[
    layers.Input((2*imsize,imsize,3)),   
    layers.Reshape((2,imsize,imsize,3)), # Splits image into 2 with 3 dimensions        
    layers.LayerNormalization(axis=[-1,-2,-3]),  

    # CNN Layers 
    layers.SeparableConv2D(32, (3, 3), activation='relu'),
    layers.MaxPooling2D(pool_size=(2, 2)),
    layers.SeparableConv2D(32, (3, 3), activation='relu'),
    layers.MaxPooling2D(pool_size=(2, 2)),
    layers.SeparableConv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D(pool_size=(2, 2)),

    layers.Flatten(),                             
    layers.Dropout(rate=0.7),                            
    layers.Dense(16,activation='relu'),           
    layers.Dense(2,activation='softmax')          
])

Though I am getting this error

Kernel shape must have the same length as input, but received kernel of shape (3, 3, 3, 32) and input of shape (None, 2, 64, 64, 3). Arguments received by SeparableConv2D.call():
     • args=('<KerasTensor shape=(None, 2, 64, 64, 3), dtype=float32, sparse=False, 
    name=keras_tensor_141>',)
     • kwargs=<class 'inspect._empty'>

I have trouble in understanding how to fit the LayerNormalization layer into SeparableConv2D. Any help on what I am doing wrongly?