Advertisement
Guest User

Untitled

a guest
Dec 9th, 2019
99
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 0.56 KB | None | 0 0
  1. OUTPUT_CHANNELS = 3
  2. base_model = tf.keras.applications.MobileNetV2(input_shape=[256, 256, 3], include_top=False)
  3.  
  4. Use the activations of these layers
  5. layer_names = [
  6. 'block_1_expand_relu', # 64x64
  7. 'block_3_expand_relu', # 32x32
  8. 'block_6_expand_relu', # 16x16
  9. 'block_13_expand_relu', # 8x8
  10. 'block_16_project', # 4x4
  11. #]
  12. #layers = [base_model.get_layer(name).output for name in layer_names]
  13.  
  14. # Create the feature extraction model
  15. #down_stack = tf.keras.Model(inputs=base_model.input, outputs=layers)
  16.  
  17. #down_stack.trainable = False
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement