将图像馈送到预训练的 keras model

Question

I trained this model (straight up copy paste from the website) and saved it with model.save() . I now want to use it to classify images that I generate so I save them and reshape them to 28x28 pixels and then attempt to feed them to the model like so:

from matplotlib import image
img = image.imread('img.png')[:,:,:1] #so that the shape ends up being (28,28,1)
print(self.model.predict(img))

But when I run this i get a bunch of errors:

WARNING:tensorflow:Model was constructed with shape (None, 28, 28, 1) for input Tensor("input_1:0", shape=(None, 28, 28, 1), dtype=float32), but it was called on an input with incompatible shape (None, 28, 1, 1). ...

ValueError: Input 0 of layer dense_12 is incompatible with the layer: expected axis -1 of input shape to have value 784 but received input with shape [None, 28]`

I've done some digging around and it seems there's an issue with the shape of the input, according to this line: WARNING:tensorflow:Model was constructed with shape (None, 28, 28, 1) for input Tensor("input_1:0", shape=(None, 28, 28, 1), dtype=float32), but it was called on an input with incompatible shape (None, 28, 1, 1)

How do I convert my image into the correct shape?

Answer 1

So i managed to fix it, embarassingly easy but i will leave this here in case anyone else has an issue:

from keras.preprocessing.image import load_img, img_to_array

img = load_img('img.png')
img = img_to_array(img)[:,:,:1]
img = np.expand_dims(img ,axis=0)

this changed the shape of my image to (1, 28, 28, 1) which is what was needed to feed to the model