model.predict() on exported model

[GaganaB]

Hi,

As I explained in this issue here: #95
I was able to resolve exporting the model and trained weights.

test_dicoms = ['test1.png','test2.png','test3.png','test4.png','test5.png','test6.png']
for img in test_dicoms: 
        full_size_image = io.imread(img, as_gray=True)
        model.predict(np.asarray(full_size_image))

throws the following error:
ValueError: Error when checking model input: the list of Numpy arrays that you are passing to your model is not the size the model expected. Expected to see 2 array(s), but instead got the following list of 1 arrays: [array([[115, 164, 173, 170, 174, 167, 148, 125, 124, 126, 150, 191, 198,
206, 205, 209, 191, 180, 128, 109, 99, 67, 63, 58, 50, 41,
32, 41],
[150, 173, 191, 188, 191, 1...

Any help is appreciated. Thanks.

Note: test_dicoms has the correct path and images are already 28 * 28 (verified with image.shape).

[XifengGuo]

@GaganaB Did you use the model or eval_model?

CapsNet-Keras/capsulenet.py

Lines 242 to 244 in 2f6d47d

	model, eval_model, manipulate_model = CapsNet(input_shape=x_train.shape[1:],
	n_class=len(np.unique(np.argmax(y_train, 1))),
	routings=args.routings)

If you use model, which is 2-in-2-out model, you should feed a tuple with 2 arrays, like outputs = model.predict(x, y).

CapsNet-Keras/capsulenet.py

Line 69 in 2f6d47d

train_model = models.Model([x, y], [out_caps, decoder(masked_by_y)])

If you use eval_model, which is a 1-in-2-out model, you should feed an array, like outputs = model.predict(x)

CapsNet-Keras/capsulenet.py

Line 70 in 2f6d47d

eval_model = models.Model(x, [out_caps, decoder(masked)])