Classifying the wrong image in federated learning duet_mnist url_image example
mohres opened this issue · comments
Mohammad Fares commented
Description
In federated-learning -> duet_mnist -> MNIST_Syft_Data_Scientist notebook, the function classify_url_image
does not really classify the received image's url, it predicts the value of image_1
which is an example from the test dataset.
How to Reproduce
- Go to 'federated-learning/duet_mnist/MNIST_Syft_Data_Scientist.ipynb'
- Scroll down to 'classify_url_image', notice that the function only shows the received image in line
plt.imshow(im, ...
, but predicts a different one in lineclass_num, preds = classify_local(image_1, local_model)
Expected Behavior
The expected behavior is that the function downloads the given image image_url
, displays it, and predicts the digit that it represents.