ValueError: Graph disconnected: cannot obtain value for tensor KerasTensor(type_spec=TensorSpec(shape=(None, 224, 224, 3)

I trained a pretrained model with vgg19 on 9 classes. I saved the model and load it to predict on images into a folder. When I try to predict, I have this error:

ValueError: Graph disconnected: cannot obtain value for tensor KerasTensor(type_spec=TensorSpec(shape=(None, 224, 224, 3), dtype=tf.float32, name='input_1'), name='input_1', description="created by layer 'input_1'") at layer "block1_conv1". The following previous layers were accessed without issue: []

I declare my pretrained model like that:

from keras.applications.vgg19 import VGG19
model_name='VGG19'
base_model = VGG19(input_shape=(224, 224, 3), weights='imagenet', pooling="avg", include_top=False)
x=base_model.output
x=keras.layers.BatchNormalization(axis=-1, momentum=0.99, epsilon=0.001 )(x)
x = Dense(256, kernel_regularizer = regularizers.l2(l = 0.016),activity_regularizer=regularizers.l1(0.006),
                bias_regularizer=regularizers.l1(0.006) ,activation='relu')(x)
x=Dropout(rate=.45, seed=123)(x)        
output=Dense(class_count, activation='softmax')(x)
model=Model(inputs=base_model.input, outputs=output)
model.compile(Adamax(learning_rate=.001), loss='categorical_crossentropy', metrics=['accuracy'])

After the training, I save the model and I load it: from tensorflow.keras.models import load_model

#save the model
model.save('model_he.h5')
#load model
m=load_model('model_he.h5')

When I try to predict on the folder with my pretrained model, the error Graph disconnected appears.

import os

import cv2
import argparse
import numpy as np
from keras.applications.vgg19 import VGG19
from keras.preprocessing import image as image_utils
from keras.applications.imagenet_utils import preprocess_input, decode_predictions

# Construct argument parser and parse the arguments
argument_parser = argparse.ArgumentParser()

# First two arguments specifies our only argument "image" with both short-/longhand versions where either 
# can be used
# This is a required argument, noted by required=True, the help gives additional info in the terminal
# if needed
argument_parser.add_argument("-i", "--image", required=True, help="path to the input image")

# Set path to files
img_path = "val/"
files = os.listdir(img_path)
print("[INFO] loading and processing images...")

for filename in files:
    
    # Passing the entire path of the image file
    file= os.path.join(img_path, filename)
    
    # Load original via OpenCV, so we can draw on it and display it on our screen
    original = cv2.imread(file)

    image = image_utils.load_img(file, target_size=(224, 224))
    image = image_utils.img_to_array(image)

    image = np.expand_dims(image, axis=0)
    image = preprocess_input(image)


    print("[INFO] loading network...")
    

    print("[INFO] classifying image...")
    predictions = m.predict(image)   # Classify the image (NumPy array with 1000 entries)
    predictions = Dense(9, activation='softmax')(x)
    model = Model(inputs=m.input, outputs=predictions)
    #P = decode_predictions(predictions)  # Get the ImageNet Unique ID of the label, along with human-readable label
    print(P)    

    # Loop over the predictions and display the rank-5 (5 epochs) predictions + probabilities to our terminal
    for (i, (imagenetID, label, prob)) in enumerate(P[0]):
        print("{}. {}: {:.2f}%".format(i + 1, label, prob * 100))

    original = cv2.imread(file)
    (imagenetID, label, prob) = P[0][0]
    cv2.putText(original, "Label: {}, {:.2f}%".format(label, prob * 100), (10, 30), 
                cv2.FONT_HERSHEY_SIMPLEX, 0.8, (0, 255, 0), 2)
    cv2.imshow(original)  
    cv2.waitKey(0) 

Any idea?