model_config = json_utils.decode(model_config.decode('utf-8')) AttributeError: 'str' object has no attribute 'decode'

Question

I am working on a project on Python that detects disease on leaves and sprays fertilizer on the leaf.

After many hours of troubleshooting other errors, I came down to the following final error that always happens and I can't seem to fix.

Following are the versions I have used for the dependencies so far:

• keras = 2.4.3
• cv2 = 4.5.1
• numpy = 1.20.1
• tensorflow = 2.4.0
• h5py = 3.2.1
• pandas = 1.2.3

Error that I am facing:

Traceback (most recent call last):
  File "leaf_cnn.py", line 12, in <module>
    model = load_model('Leaf_CNN.h5')
  File "/home/pi/.local/lib/python3.7/site-packages/tensorflow/python/keras/saving/save.py", line 207, in load_model
    compile)
  File "/home/pi/.local/lib/python3.7/site-packages/tensorflow/python/keras/saving/hdf5_format.py", line 182, in load_model_from_hdf5
    model_config = json_utils.decode(model_config.decode('utf-8'))
AttributeError: 'str' object has no attribute 'decode'

Code file leaf_cnn.py

# importing files/dependencies

from tensorflow.keras.models import load_model
import cv2
import numpy as np
#import Categories
import time
import RPi.GPIO as GPIO

#loading model/ML algorithm 

model = load_model('Leaf_CNN.h5')
cap = cv2.VideoCapture(0) # capture frame
ret, img = cap.read()

channel = 21 
GPIO.setmode(GPIO.BCM)
GPIO.setup(channel,GPIO.OUT)
#cv2.imshow('aaa',img)  'display image with title'

img = cv2.resize(img,(224,224)) 
img = np.reshape(img,[1,224,224,3])
classes = model.predict(img)
y_pred = np.argmax(classes, axis=1)
y_pred = Categories.categories[int(y_pred)]

if "healthy" not in y_pred:
    GPIO.output(21, GPIO.HIGH) #turn-on relay
    time.sleep(1)
else:
    GPIO.output(21, GPIO.LOW) #turn-off relay
    time.sleep(1)
#cv2.waitKey(0)
#cv2.destroyAllWindows()

Answer 1

There seems to be an issue with h5py and it has to be installed in a certain version to ensure compatibility with tensorflow. This is what did the trick for me:

pip uninstall h5py
pip install h5py==2.10.0