Different results from model.predict() and model.predict_generator()

I am getting different results for the prediction of the same image when I use model.predict and when I use model.predict_generator

Here is how I load the image that I am passing to model.predict
x = image.load_img(image_file_path, target_size=(224, 224))
x = image.img_to_array(x)
x = np.expand_dims(x, axis=0)

this is an issue that i have encountered before (i think it is a known issue). just use ‘predict’ and normalize your data without a generator (if applicable)