[Errno 13] Permission denied

[secco88]

Hello everyone, I'm trying to use the custom train but I get this error:

PermissionError: [Errno 13] Permission denied: 'C:\Users\secco\Desktop\ML\olivi\train.json'

I think the problem is that the script is trying to open this folder as a JSON file.

import pixellib
from pixellib.custom_train import instance_custom_training
vis_img = instance_custom_training()
vis_img.load_dataset("olivi")

[greg007-web]

The new version didn't work. I had the same problem. Try this setting:

!pip3 install tensorflow==2.4.1
!pip3 install tensorflow--gpu
!pip3 install imgaug
!pip install pixellib  --upgrade
!pip install labelme2coco==0.1.2

I used an old version and now is working.

[secco88]

Thanks I have do a similar configuration with Anaconda:

python 3.7
conda install tensorflow-gpu==2.5.0
pip install labelme2coco==v0.1.2
pip install -U scikit-image==0.16.2
pip install imgaug==0.4.0

[Tarandeep97]

I am still facing this issue. Can you please specify in which setting you are using, jupyter notebook, colab?

The new version didn't work. I had the same problem. Try this setting:

!pip3 install tensorflow==2.4.1
!pip3 install tensorflow--gpu
!pip3 install imgaug
!pip install pixellib  --upgrade
!pip install labelme2coco==0.1.2

I used an old version and now is working.

[greg007-web]

I used this on Google Colab Em ter., 29 de mar. de 2022 14:34, Tarandeep Singh ***@***.***> escreveu:

…

I am still facing this issue. Can you please specify in which setting you are using, jupyter notebook, colab? The new version didn't work. I had the same problem. Try this setting: !pip3 install tensorflow==2.4.1 !pip3 install tensorflow--gpu !pip3 install imgaug !pip install pixellib --upgrade !pip install labelme2coco==0.1.2 I used an old version and now is working. — Reply to this email directly, view it on GitHub <#123 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ARRC2PFLJ4IZR7MVE6YN5HDVCM5KFANCNFSM5NAQ22FQ> . You are receiving this because you commented.Message ID: ***@***.***>

[mjan2021]

The new version didn't work. I had the same problem. Try this setting:

!pip3 install tensorflow==2.4.1
!pip3 install tensorflow--gpu
!pip3 install imgaug
!pip install pixellib  --upgrade
!pip install labelme2coco==0.1.2

I used an old version and now is working.

Hi
Can you specify the version you used that is working.

Thanks

[palvors]

So, I Found, the labelme2coco package create a problem.

presently that create a folder train.json and file train.json...

and the ini.py of this package you will find def convert.

I just replace

save_path = str(Path(export_dir) / "dataset.json")

by

save_path = export_dir

so result :

Converting labelme annotations to COCO format: 100%|██████████| 600/600 [00:00<00:00, 1960.44it/s]
09/13/2022 23:13:52 - INFO - labelme2coco - Converted annotations in COCO format is exported to e:\Projetsa\test\pixellib\Nature\train.json

There are 200 listed files in folder test.
Converting labelme annotations to COCO format: 100%|██████████| 200/200 [00:00<00:00, 2134.51it/s]
09/13/2022 23:13:53 - INFO - labelme2coco - Converted annotations in COCO format is exported to e:\Projets\a\test\pixellib\Nature\test.json

And after, let the code train

Applying Default Augmentation on Dataset
Train 0 images
Validate 0 images
Checkpoint Path: e:\Projetsa\test\pixellib\mask_rcnn_models
Selecting layers to train
Epoch 1/10