TensorFlow Checkpoint Path Issue in MATLAB-Python Integration

[braahkrayem]

I'm encountering an issue when running the orig_mod_NNEval.py script from MATLAB. The script works correctly when executed manually from the Command Prompt (CMD), and the .mat file (CNN_striatal_python_output_intermediate.mat) is generated successfully. However, when executed via MATLAB using the main_CNNStriatalSegmentation function, the TensorFlow checkpoint path becomes corrupted, and the script fails to load the model.

[yashpat016]

Hi @braahkrayem

Thank you for reporting the issue with running the orig_mod_NNEval.py script through MATLAB. To better understand and resolve the problem, could you please provide some additional details?

1. Python Environment:

• Are you running the script from a virtual environment when using cmd?
• Could you confirm if the virtual environment is activated when running the script through MATLAB?

2. Error Logs:

• Could you share the complete error log when you attempt to run the script via MATLAB? This would help in pinpointing or replicating the exact step where the issue occurs.

I’ve attempted to replicate this error and the closest I came was encountering an import error when calling the script from outside the virtual environment when running the main_CNNStriatalSegmentation function. I believe that the issue may be related to how MATLAB interfaces with the virtual environment, particularly with respect to locating and using the necessary Python libraries.

Could you confirm if your setup is similar, and if so, how you typically manage the integration of your virtual environment with MATLAB?

Regards,
Yash

[braahkrayem]

- Thank you for your prompt response and willingness to assist with this issue. Below are the details you requested: - I am *not* running the script from a virtual environment when executing it via CMD. - n MATLAB, I am pointing directly to the Python executable located at: makefile Copy code C:\Users\ACER\AppData\Local\Programs\Python\Python311\python.exe *Python Libraries:* - Python Version: *3.11.6* - TensorFlow Version: *2.10.0* - NumPy Version: *1.23.2* - SciPy Version: *1.9.0* - Nibabel Version: *4.0.2* - * When I run the script via MATLAB, I encounter the following error:plaintextCopy codeWARNING:tensorflow:ParseError: 1:24 : 'model_checkpoint_path: "C:': String missing ending quote: '"C:' WARNING:tensorflow:C:\Users\ACER\OneDrive\Desktop\braah_project\scripts\StriatalSegmentation\checkpoint: Checkpoint ignored Traceback (most recent call last): File "C:\Users\ACER\OneDrive\Desktop\braah_project\scripts\orig_mod_NNEval.py", line 97, in <module> new_saver.restore(sess, tf.train.latest_checkpoint(sys.argv[3])) File "C:\Users\ACER\AppData\Local\Programs\Python\Python311\Lib\site-packages\tensorflow\python\training\saver.py", line 1406, in restore raise ValueError("Can't load save_path when it is None.") ValueError: Can't load save_path when it is None. * - The checkpoint file (checkpoint) inside the StriatalSegmentation directory seems to get altered or corrupted when called from MATLAB. - Manually running the script via CMD with the same arguments produces no errors, and the checkpoint path is correctly recognized. *Best regards,* *Braah krayem*

…

On Thu, Dec 26, 2024 at 7:52 PM yashpat016 ***@***.***> wrote: Hi @braahkrayem <https://github.com/braahkrayem> Thank you for reporting the issue with running the orig_mod_NNEval.py script through MATLAB. To better understand and resolve the problem, could you please provide some additional details? 1. Python Environment: - Are you running the script from a virtual environment when using cmd? - Could you confirm if the virtual environment is activated when running the script through MATLAB? 2. Error Logs: - Could you share the complete error log when you attempt to run the script via MATLAB? This would help in pinpointing or replicating the exact step where the issue occurs. I’ve attempted to replicate this error and the closest I came was encountering an import error when calling the script from outside the virtual environment when running the main_CNNStriatalSegmentation function. I believe that the issue may be related to how MATLAB interfaces with the virtual environment, particularly with respect to locating and using the necessary Python libraries. Could you confirm if your setup is similar, and if so, how you typically manage the integration of your virtual environment with MATLAB? Regards, Yash — Reply to this email directly, view it on GitHub <#4 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BNLBX4BOROCMQXYWANTGSCT2HQ65VAVCNFSM6AAAAABUGKEGNGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNRSHE4DONJQGM> . You are receiving this because you were mentioned.Message ID: <CNaP-Lab/Toolkit_for_MIRA_LAB_Striatal_Segmentation/issues/4/2562987503@ github.com>