Unable to see any ouput

[kkarnatak]

I am trying to run a simple example using trained network vgg16

from vgg16 import VGG16
from quiver.quiver_engine.server import launch

model = VGG16()
launch(model, classes=['cat','dog'], input_folder='./imgs') # i have images of dogs and cats here

After this I can see the webpage, but when I click on image and then the layer, it displays "no data for this layer". And this is displayed for all the layers.

I can see following error on the console:

  File "C:\ENV\p34\lib\site-packages\keras\engine\training.py", line 111, in standardize_input_data
    str(array.shape))
Exception: Error when checking : expected input_1 to have shape (None, 224, 224, 3) but got array with shape (1, 224, 3, 3)

I used 224x224 image as an input. What might I be missing here?

[jakebian]

That error says apparently you are attempting to evaluate this on 1 image of size 224 x 3 px. Could be a preprocessing error though - can you give some more details on your input images? How many images are in the folder, what's the file extension?

[kkarnatak]

Thanks for replying. At the moment I am using just one image with extension jpg. Its dimensions are 224x224 and 24bit. I believe 224x224 is the required size, and I tried to feed the images with same dimensions but every images leads to the error stated in the post above. CAn you recommened any images which have worked on this code?

[MikhailovSergei]

Hello, do solve this problem?

[sudheerExperiments]

I am facing similar issue. My model uses 1 x 120 x 120 (channels x width x height) black and white images. My dataset has 300 x 300 (channels x width x height) black and white images. I resized the images to 120x120 while feeding data to my model.

When I used quiver to display conv layer outputs, input loads, but shows 'No data for this layer' error.

Any solution??

[rajee-a]

Successfully loaded but not output ?