hulianyuyy / STGAT

Skeleton-Based Action Recognition with Local Dynamic Spatial-Temporal Aggregation (Expert Systems with Applications 2023) (Previous name: Spatial Temporal Graph Attention Network for Skeleton-Based Action Recognition)
Other
37 stars 8 forks source link

Not using gpu #15

Closed mahshidaj95 closed 1 year ago

mahshidaj95 commented 1 year ago

Hi, I'm using your code, but I wonder if is it possible to not use gpu and run the code on copu? My NVIDIA version is too old so I can't use cuda gpu.

hulianyuyy commented 1 year ago

Currently, this code doesn't support cpu. And i infer that running this code on cpu may be too slow. Running it on a 2080ti GPU costs about one day. It may consume more than a week on a common cpu.

mahshidaj95 commented 1 year ago

That's right. thank you.

mahshidaj95 commented 1 year ago

Hello Dear Hulianyuyy, I want to run an attention model on skeleton dataset similar to your github code (STGAT) for human action recognition. However, I don't have a GPU, so I want to run on a CPU server. Do you know what the minimum RAM requirement I need?

Sincerely, Mahshid Alijabbari

On Sun, Jun 25, 2023 at 6:21 AM hulianyu @.***> wrote:

Currently, this code doesn't support cpu. And i infer that running this code on cpu may be too slow. Running it on a 2080ti GPU costs about one day. It may consume more than a week on a common cpu.

— Reply to this email directly, view it on GitHub https://github.com/hulianyuyy/STGAT/issues/15#issuecomment-1606091255, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIJ2PCVTLMY45CRB6EB7A3DXNA3OFANCNFSM6AAAAAAZTEYZWY . You are receiving this because you authored the thread.Message ID: @.***>

hulianyuyy commented 1 year ago

I'm sorry that i haven't run this code on the cpu server. Thus i don't know the minimum RAM requirement. Maybe you can find a repo that supports the cpu server which may better help you.

mahshidaj95 commented 1 year ago

Thank you for your response. Actually, my thesis is human action recognition using graph attention network and I try many codes and neural network models using attention mechanism, but I didn't get appropriate results. Can you send me a simpler attention model to fit to the NTU skeleton dataset? I would be so grateful.

On Sun, Jul 16, 2023 at 5:37 AM hulianyu @.***> wrote:

I'm sorry that i haven't run this code on the cpu server. Thus i don't know the minimum RAM requirement. Maybe you can find a repo that supports the cpu server which may better help you.

— Reply to this email directly, view it on GitHub https://github.com/hulianyuyy/STGAT/issues/15#issuecomment-1637073841, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIJ2PCTYP7P64RCHDBC4DLLXQPN7LANCNFSM6AAAAAAZTEYZWY . You are receiving this because you modified the open/close state.Message ID: @.***>