Tensorflow 1.15 Installation Problem in python 3.8+ version

[Dhyani01]

If your python version is 3.7 or less this requirement.txt installation will work great,
Those who have python version 3.8+ will face some issues while installing Keras 1.15

For Python version 3.8+ first, install TensorFlow by

pip install tensorFlow

it will install TensorFlow version 2.5
Do these following steps in order to make the project working :

1. Replace this line graph = tf.get_default_graph() by graph= tf.compat.v1.get_default_graph() in deeplearning.py
2. Remove this import,

from keras.models import load_model

by this.

from tensorflow import keras

3. Model is loaded in deeplearning.py instead load it inside the working code

with graph.as_default():
model = tf.keras.models.load_model('AlexNetModel.hdf5')
prediction = model.predict(image_array)

4: The code is using Keras library for importing images you can now import it by Image library.

comment this line# img = image.load_img('sample_image.png', target_size=(224, 224))
comment this line# img = image.img_to_array(img)
comment this line# img = np.expand_dims(img, axis=0)
comment this line# img = img/255
comment this line# print(type(img))
comment this line# img = tf.keras.preprocessing.image.array_to_img(img_data)

Now replace those lines by these lines :

input_image=Image.open("sample_image.png")
input_image = input_image.resize((224,224), resample=Image.LANCZOS)
image_array=np.array(input_image)
image_array=image_array.astype('float32')
image_array/=255
image_array=image_array.reshape(1,224,224,3)

These steps will fix your code according to the version.

[garudaonekh]

Hello,
Can you provide more detail on step4? I got stuck when I upload image, it errors with Image library.

[Dhyani01]

@garudaonekh If you have done all the steps from 1 to 3 correctly. Then for step 4 go to view.py inside plant_app folder and replace those lines above with the new ones.

[ramyaperumandla24]

@Dhyani01 @saroz014 @garudaonekh

I am facing this issue after clicking on predict button, can u please help with this?