How can I print fllaten Layers Output in CNN

Question

I use a two path CNN for image processing. I want to print the Output of flatten layer. but when I want to print the output of flatten layer i am encountering this error:

None Traceback (most recent call last):

extracted_features = model.layers32

File "C:\Miniconda3\envs\py36\lib\site-packages\keras\engine\base_layer.py", line 440, in call self.assert_input_compatibility(inputs)

File "C:\Miniconda3\envs\py36\lib\site-packages\keras\engine\base_layer.py", line 285, in assert_input_compatibility str(inputs) + '. All inputs to the layer '

ValueError: Layer flatten_39 was called with an input that isn't a symbolic tensor. Received type: <class 'numpy.ndarray'>. Full input: [array([[[[0.51611316, 0.39746055, 0.5936753 ], [0.13327277, 0.34352553, 0.7586124 ], [0.99989164, 0.43303576, 0.5733409 ], ..., [0.05266873, 0.37530503, 0.12953256], [0.2471831 , 0.71032923, 0.25836825], [0.90023595, 0.96091986, 0.3706109 ]],

    [[0.14299522, 0.1676571 , 0.55554783],
     [0.01679909, 0.32093376, 0.8943776 ],
     [0.53657365, 0.83452356, 0.1076134 ],
     ...,
     [0.77388775, 0.6101447 , 0.14687575],
     [0.47331956, 0.5162232 , 0.5501466 ],
     [0.8316039 , 0.12746046, 0.23557535]],

    [[0.745321  , 0.12560171, 0.43350193],
     [0.7007471 , 0.5507861 , 0.19451097],
     [0.55038804, 0.05661515, 0.03471742],
     ...,
     [0.70201683, 0.1263634 , 0.13524196],
     [0.24133243, 0.564834  , 0.64897084],
     [0.15477973, 0.9518493 , 0.5475725 ]],

    ...,

    [[0.8317609 , 0.5864187 , 0.48500198],
     [0.14422454, 0.05594445, 0.5835567 ],
     [0.67728573, 0.63966084, 0.18180694],
     ...,
     [0.5717842 , 0.41698074, 0.22059603],
     [0.9030291 , 0.51441693, 0.2996456 ],
     [0.9422357 , 0.03952885, 0.5567205 ]],

    [[0.28786188, 0.4799473 , 0.92904633],
     [0.08306517, 0.9738849 , 0.9621486 ],
     [0.06010618, 0.47993308, 0.5765873 ],
     ...,
     [0.8719943 , 0.43593037, 0.53257006],
     [0.11550295, 0.48797372, 0.66570866],
     [0.32191348, 0.88470393, 0.19814569]],

    [[0.39397806, 0.78230894, 0.41833642],
     [0.38489413, 0.43042594, 0.76435673],
     [0.94453263, 0.5084677 , 0.8752447 ],
     ...,
     [0.16410913, 0.45611975, 0.23094064],
     [0.6859947 , 0.57103187, 0.3729119 ],
     [0.7035603 , 0.72346485, 0.63314646]]]], dtype=float32)]. All inputs to the layer should be tensors.

Code:


input_shape = (IMAGE_WIDTH, IMAGE_HEIGHT, 3)

num_classes = 3

# Data
path = "D:/Datesets/4- Card/data"
modelName = '2_Class'
training_data_dir = path + "/training" # 500 * 2
validation_data_dir = path + "/validation" # 100 * 2
test_data_dir = path + "/test" # 150
MODEL_FILE = modelName+ '-Vgg&ResnetBachSize'+ str(BATCH_SIZE)+ 'Epoch'+ str(EPOCHS)+ '.h5'

input1 = modelBuilder.defineInput()
input2 = modelBuilder.defineInput()

# First Path Of Training
finput1=modelBuilder.configInitialModel(input1)           finput2=modelBuilder.configInitialModel(input2)  
                                                          
output=modelBuilder.concatinateflattenInputs(finput1,finput2)

model=modelBuilder.makeModel(input1, input2, output) #(input1, input2, output)
print(model.summary())

img = np.random.random((1, 200, 200, 3)).astype("float32")

extracted_features = model.layers[32](img)

extracted_features = model.layers[33](img)

sanyam · Answer 1 · 2022-09-21T05:22:48.380

1

I think you are facing this issue because you are passing a numpy array to the layer (check the error message). Could you try transforming the numpy array to a tensor?

Something like

img = torch.tensor(img)
extracted_features = model.layers[32](img)

If you are working with keras, the step would look like this

Edit Copied from the answer above -

For keras/tensorflow, try this

import tensor as tf
#Considering y variable holds numpy array
y_tensor = tf.convert_to_tensor(y, dtype=tf.int64)

edited Sep 21 '22 at 05:22

answered Sep 20 '22 at 17:06

sanyam

76
1
3

img = tf.tensor(img) > AttributeError: module 'tensorflow' has no attribute 'tensor' – Amir Sep 20 '22 at 17:09
`import tensor as tf` `#Considering y variable holds numpy array` `y_tensor = tf.convert_to_tensor(y, dtype=tf.int64)` Try this – sanyam Sep 21 '22 at 05:18
I used the following command to display and print the output of the felt layer. However, it is not possible to display and print the output of this layer and >Tensor("flatten_1_1/Reshape:0", shape=(1, 120000), dtype=int64) is displayed. I want to print values of this Layer – Amir Sep 25 '22 at 05:58
So is it working now? That's good. What is happening is that the you are receiving the object name when you're trying to print the value. I'd suggest you to read the documentation and go through [this answer](https://stackoverflow.com/questions/33633370/how-to-print-the-value-of-a-tensor-object-in-tensorflow) for the solution. – sanyam Sep 27 '22 at 12:52

How can I print fllaten Layers Output in CNN

1 Answers1