FileNotFoundError: [Errno 2] No such file or directory: './models/kinetics-st_gcn.pt'

[manza-ari]

@yysijie

Code version (Git Hash) and PyTorch version

0.4.0

Dataset used

kinetics-skeleton

Expected behavior

I have just started to implement the project, tried to test or train or see the demo.

Actual behavior

I have just started to implement the project, tried to test or train or see the demo for all three it says:
FileNotFoundError: [Errno 2] No such file or directory: './models/kinetics-st_gcn.pt'

Steps to reproduce the behavior

Other comments

[yysijie]

Hi,
Did you run bash tools/get_models.sh successfully?

[manza-ari]

Thank you so much for your reply!

Yes, I ran it successfully and run again after reading your message. Check the attached screenshot.

[yysijie]

Hi, @kanza-ali
can you show me the command you ran and more detail about error messages?

[manza-ari]

@yysijie Thank you for your reply.

I think the issue is resolved but I have some new errors.

For Demo I am running these commands

python3 main.py demo
--openpose '/home/kanza/opencv-3.1/opencv-3.1.0/build'
--video '/home/kanza/Downloads/st-gcn-master/v_TableTennisShot_g02_c01.avi'
--device 0

For Training

python3 main.py recognition -c config/st_gcn/kinetics-skeleton/train.yaml --device 0

[yysijie]

@kanza-ali ,
For running the demo, I think you should build and install the Openpose first.
For training, your GPU memory is not enough for default hyper-parameters. You can use a smaller batch size to lower memory requirement. Just add --batch_size <int> and --test_batch_size <int> in the command line. Such as:

python3 main.py demo
--openpose '/home/kanza/opencv-3.1/opencv-3.1.0/build'
--video '/home/kanza/Downloads/st-gcn-master/v_TableTennisShot_g02_c01.avi'
--device 0
--batch_size 16
--test_batch_size 16

[manza-ari]

thank you so much @yysijie for your reply.
You are right, I will definitely follow your solutions.