Running the code with GPU

[DaneshShokri94]

Dear

First of all, I appreciate your work, that's awsome

I am running masa on the anaconda - windows. The code use almost all 32G RAM of my computer. it took around 45 minutes for analysing a 90 second video.

I would like to use 8g GPU of my computer for accelerating the computation time

Could you please help me how can I do this?

Sincerely

[siyuanliii]

Thanks for your question! If you run the default GoundingDINO model with a SwinB backbone on a CPU, it can be very inefficient. Our code, by default, supports inference with GPU. Make sure you have the GPU correctly installed and select the corresponding inference device.

[DaneshShokri94]

Dear Siyuanliii

Thanks for your respone. Actually, I tried my best to run the code under GPU, but it is not working unforetunatelly. All of needed steps to install GPU on my Anaconda (Windows version) have been done correctly.

Sincerely

[siyuanliii]

I need more information to help you with this question. Have you installed the correct torch version for GPU? Have you done the check, for example, whether your GPU is available by torch.cuda.is_available()

[DaneshShokri94]

Dear @siyuanliii

Thank you for your response, After running this code "torch.cuda.is_available()", I get True. Also by running this code "torch.cuda.get_device_name()", I get the name of my gpu "NVIDIA RTX A2000 8GB Laptop GPU"

When I run the following code from your code "python demo/video_demo_with_text.py demo/driving_10s.mp4 --out demo_outputs/driving_10s_outputs.mp4 --det_config projects/CO-DETR/configs/codino/co_dino_5scale_r50_lsj_8xb2_3x_coco.py --det_checkpoint saved_models/pretrain_weights/co_dino_5scale_lsj_r50_3x_coco-fe5a6829.pth --masa_config configs/masa-one/masa_r50_plug_and_play.py --masa_checkpoint saved_models/masa_models/masa_r50.pth --score-thr 0.3 --show_fps"

The code is run like the below image on my Anaconda (Windows version). some warnings are appears. maybe this wrning are the cause of this gpu issue.

[RecardoLyu]

Hi @DaneshShokri94
I'm trying ro run the demo, too. But I'm facing anothor issue, that is, some packages used in the code (import resource) is only effective in linux environment. It seems you could run the demo, have you encountered this problem? If yes, how did you solve the problem?
Sincerely

Dear @siyuanliii

Thank you for your response, After running this code "torch.cuda.is_available()", I get True. Also by running this code "torch.cuda.get_device_name()", I get the name of my gpu "NVIDIA RTX A2000 8GB Laptop GPU"

When I run the following code from your code "python demo/video_demo_with_text.py demo/driving_10s.mp4 --out demo_outputs/driving_10s_outputs.mp4 --det_config projects/CO-DETR/configs/codino/co_dino_5scale_r50_lsj_8xb2_3x_coco.py --det_checkpoint saved_models/pretrain_weights/co_dino_5scale_lsj_r50_3x_coco-fe5a6829.pth --masa_config configs/masa-one/masa_r50_plug_and_play.py --masa_checkpoint saved_models/masa_models/masa_r50.pth --score-thr 0.3 --show_fps"

The code is run like the below image on my Anaconda (Windows version). some warnings are appears. maybe this wrning are the cause of this gpu issue.

[DaneshShokri94]

Dear @RecardoLyu

Simply you can solve this problem.

Firstly open the code of "video_demo_with_text.py" at the folder of "demo".

Next, delete or comment the lines code of 8, 37,38,39,40,41,42 (actually this is a function).