CODEWITHSUNDEEP
python-3.xmachine-learningtf.keras
Udit Choudhury
Udit Choudhury

Reputation: 11

How to print the accuracy of the prediction done by a saved model?

I trained a CNN to categorize between dogs and cats and saved it as: MiniProject_01.hdf5 with accuracy of 80%

Now I want to test my model. Here is my code:

import cv2
import keras

CATEGORIES =["Dog", "Cat"]

def data(file):
    IMG_SIZE = 64
    img_array = cv2.imread(file, cv2.IMREAD_GRAYSCALE)
    new_array = cv2.resize(img_array, (IMG_SIZE, IMG_SIZE))
    return new_array.reshape(-1, IMG_SIZE, IMG_SIZE, 1)

model = keras.models.load_model("MiniProject_01.hdf5")

prediction = model.predict([data("dog13.jpg")])
print(CATEGORIES[int(prediction[:])])

It just prints: Dog.

Can I get the percent accuracy/probability by which my model said it is a dog?

Upvotes: 0

Views: 868

Answers (2)

Playergod
Playergod

Reputation: 119

model.predict() returns a 2dim array with the probability for each output.

So try it with prediction[0][0] for the Dog or prediction[0][1] for the Cat.

Upvotes: 0

W A Geraldine
W A Geraldine

Reputation: 11

Maybe you mean is percentage confidence. You can use prediction = model.predict_proba([data("dog13.jpg")]) to get it.

Upvotes: 1

Related Questions