Kernel of jupyter notebook die when calling vgg16 model predict

dvd · May 31, 2023, 3:02pm

Hello,

i can detect and use my GPU card if i use tensorflow bert, or tensorflow model.sequential for example ,

i can reproduce in my jupyter notebook the code lines in the website
https://machinelearningmastery.com/tutorial-first-neural-network-python-keras/

but if i try to use tensorflow keras VGG16 (pretrained model) when i use model.predict, the kernel die.

I hope someone may help me to resolve this problem in order to use VGG16 for a classification of images.

i listed below all the information i was able to find

the code, and the type of error,
the list of packages in my conda environment,
my system and the list of programs installed (microsoft visual ++, python, anaconda, nvidia)
my jupyter notebook version and a verif for the installation of cuda

sincerely yours,
Dvd

about the issue with keras model VGG16 and VGG19

for downloading the picture

https://hips.hearstapps.com/hmg-prod/images/little-cute-maltipoo-puppy-royalty-free-image-1652926025.jpg?crop=0.444xw:1.00xh

import tensorflow as tf
from tensorflow.keras.models import Model, Sequential
from tensorflow.keras.applications.vgg16 import VGG16
import numpy as np

with hyperparameter include_top = False, the problem still remind the same

base_model = VGG16(weights=‘imagenet’, include_top=True)

from tensorflow.keras.preprocessing.image import load_img, img_to_array
from keras.preprocessing import image
from tensorflow.keras.applications.vgg16 import preprocess_input, decode_predictions

img_path = “./little-cute-maltipoo-puppy-royalty-free-image-1652926025.jpg”
img = load_img(img_path, target_size=(224, 224))
x = img_to_array(img)
x = np.expand_dims(x, axis=0)
x = preprocess_input(x)

print(tf.version)

the problem occurs at this step

preds = base_model.predict(x)
decode_predictions(preds, top = 3)

Error message : “Kernel restarting” “The kernel appears to have died. It will restart automatically.”

about my conda env

(mygpuenv) ./raoul>conda list

packages in environment at ./anaconda3\envs\mygpuenv:

about my system and GPU card

Système d’exploitation
-----> Windows 10 Famille 64-bit
Processeur
Intel Core i7 8750H @ 2.20GHz
RAM
16,0 Go Canal-Simple Inconnu @ 1330 MHz (19-19-19-43)
Carte mère
ASUSTeK COMPUTER INC. FX504GD (U3E1)
Graphiques
Intel UHD Graphics 630 (ASUStek Computer Inc)
----> NVIDIA GeForce GTX 1050 (ASUStek Computer Inc)

about the list of programs required for tensorflow install

Anaconda3 2023.03-1 (Python 3.10.9 64-bit)	Anaconda, Inc.	2023.03-1
Microsoft ASP.NET Core 6.0.13 - Shared Framework (x64)	Microsoft Corporation	6.0.13.22580
Microsoft Update Health Tools	Microsoft Corporation	3.72.0.0
Microsoft Visual C++ 2005 Redistributable	Microsoft Corporation	8.0.56336
Microsoft Visual C++ 2008 Redistributable - x64 9.0.30729.17	Microsoft Corporation	9.0.30729
Microsoft Visual C++ 2008 Redistributable - x64 9.0.30729.6161	Microsoft Corporation	9.0.30729.6161
Microsoft Visual C++ 2008 Redistributable - x86 9.0.30729.6161	Microsoft Corporation	9.0.30729.6161
Microsoft Visual C++ 2010 x64 Redistributable - 10.0.40219	Microsoft Corporation	10.0.40219
Microsoft Visual C++ 2010 x86 Redistributable - 10.0.40219	Microsoft Corporation	10.0.40219
Microsoft Visual C++ 2012 Redistributable (x64) - 11.0.61030	Microsoft Corporation	11.0.61030.0
Microsoft Visual C++ 2012 Redistributable (x86) - 11.0.61030	Microsoft Corporation	11.0.61030.0
Microsoft Visual C++ 2013 Redistributable (x64) - 12.0.30501	Microsoft Corporation	12.0.30501.0
Microsoft Visual C++ 2013 Redistributable (x86) - 12.0.30501	Microsoft Corporation	12.0.30501.0
Microsoft Visual C++ 2015-2022 Redistributable (x64) - 14.36.32532	Microsoft Corporation	14.36.32532.0
Microsoft Visual C++ 2015-2022 Redistributable (x86) - 14.36.32532	Microsoft Corporation	14.36.32532.0
Microsoft Windows Desktop Runtime - 6.0.13 (x64)	Microsoft Corporation	6.0.13.32001
NVIDIA Control Panel	NVIDIA Corp.	8.1.964.0
NVIDIA CUDA Development 11.2	NVIDIA Corporation	11.2
NVIDIA CUDA Documentation 11.2	NVIDIA Corporation	11.2
NVIDIA CUDA Nsight NVTX 11.2	NVIDIA Corporation	11.2
NVIDIA CUDA Runtime 11.2	NVIDIA Corporation	11.2
NVIDIA CUDA Samples 11.2	NVIDIA Corporation	11.2
NVIDIA CUDA Visual Studio Integration 11.2	NVIDIA Corporation	11.2
NVIDIA GeForce Experience 3.27.0.112	NVIDIA Corporation	3.27.0.112
NVIDIA Logiciel système PhysX 9.21.0713	NVIDIA Corporation	9.21.0713
NVIDIA Nsight Compute 2020.3.0	NVIDIA Corporation	20.3.0.0
NVIDIA Nsight Systems 2020.4.3	NVIDIA Corporation	20.4.3.7
NVIDIA Nsight Visual Studio Edition 2020.3.0.20315	NVIDIA Corporation	20.3.0.20315
NVIDIA Pilote graphique 531.61	NVIDIA Corporation	531.61
NVIDIA Tools Extension SDK (NVTX) - 64 bit	NVIDIA Corporation	1.00.00.00
PyCharm Community Edition 2022.2.3	JetBrains s.r.o.	222.4345.23
Python 3.10.11 (64-bit)	Python Software Foundation	3.10.11150.0
Python Launcher	Python Software Foundation	3.10.11150.0
Shared Add-in Support Update for Microsoft .NET Framework 2.0 (KB908002)	Microsoft	1.0.0
Webp Image Extensions	Microsoft Corporation	1.0.52351.0
Windows Print
Windows SDK AddOn	Microsoft Corporation	10.1.0.0

about jupyter notebook

Server Information:

You are using Jupyter Notebook.

The version of the notebook server is: 6.5.2
The server is running on this version of Python:

Python 3.10.9 | packaged by Anaconda, Inc. | (main, Mar 1 2023, 18:18:15) [MSC v.1916 64 bit (AMD64)]

Current Kernel Information:

Python 3.9.0 (default, Nov 15 2020, 08:30:55) [MSC v.1916 64 bit (AMD64)] Type ‘copyright’, ‘credits’ or ‘license’ for more information IPython 7.34.0 – An enhanced Interactive Python. Type ‘?’ for help.

about nvidia cuda confirmation and cmd windows

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2020 NVIDIA Corporation
Built on Mon_Nov_30_19:15:10_Pacific_Standard_Time_2020
Cuda compilation tools, release 11.2, V11.2.67
Build cuda_11.2.r11.2/compiler.29373293_0

Renu_Patel · August 17, 2023, 2:52pm

Hi @dvd

Welcome to the Tensorflow Forum!

It seems the installed Python version and the Tensorflow version are not compatible to each other which could be the reason for this issue. Python 3.10 supports Tensorflow >=2.8 as mentioned in this tested build configuration.

Could you please try again by creating a new Tensorflow environment? You can refer to this TF install page and follow the step by step instructions given in the link.

Note:

Please make sure to install Tensorflow <=2.10 because TensorFlow 2.10 was the last TensorFlow release that supported GPU on native-Windows. Starting with TensorFlow 2.11 , you will need to install TensorFlow in WSL2 or install tensorflow or tensorflow-cpu and, optionally, try the TensorFlow-DirectML-Plugin

Let us know if the issue still persists. Thank you.

dvd · January 31, 2024, 9:53pm

Hello Renu_Patel, thanks a lot for answering me, and i apologize for writing so late.

Indeed, i had to update tensorflow, it works with the version 2.12 with python 3.9.0. I don’t think this configuration allows to detect my gpu card with windows, but at least, i was in measure to call my CNN models and execute my cells, so…

i will test with pleasure your solutions when i will finish my distance training in data science field; the time is missing me for the projects for now.

Thanks a lot for your solutions,
best,
David

Topic		Replies	Views
The kernel appears to have died. It will restart automatically Keras models , datasets	7	3329	June 6, 2024
CNN network causes kernel die General Discussion wsl2 , keras	20	2299	June 6, 2024
Entire computer crashes when running tensorflow code TensorFlow models , tensorflow	2	67	November 16, 2024
Jupyter kernal crash in vscode while using tensorflow GPU Keras gpu , tfcore	1	107	July 1, 2024
The kernel appears to have died. It will restart automatically. whenever i try to run the plt.imshow() and plt.show() function in jupyter notebook Keras keras	2	334	May 12, 2024