Keras: Interpretation of model.predict() for image classification

Question

I have created an image classifier using keras and I am trying to use model.predict() for a single image. My code looks like this:

path = 's.jpg'

from keras.preprocessing import image

img_width, img_height = 224,224

img = image.load_img(path, target_size = (img_width, img_height))

img = image.img_to_array(img)

img = np.expand_dims(img, axis = 0)

a =model.predict(img)

But the output array 'a' with probability scores is always in descending order. Do you have any idea why?? I have three categories.

Answer 1

The predict returns class probability, [class1_prob, class2_prob, ....] assuming you have softmax in your model.

It should not be always decreasing unless you pass such data so that for which the class probability is always in decreasing order.