Error when running video_demo.py

Question

Error when running video_demo.py

mariosconsta opened this issue a year ago · comments

Marios Constantinou commented a year ago

I am running on google colab with GPU enabled. I downloaded the weights and have a video ready to play with. When I run the specified command, I get this error:

"RuntimeError: Attempting to deserialize object on CUDA device 1 but torch.cuda.device_count() is 1. Please use torch.load with map_location to map your storages to an existing device."

Can somebody help me? I do have a GPU ready, I do not know why it doesn't run.

Marios Constantinou · Answer 1 · Sat Mar 18 2023 01:55:59 GMT+0800 (China Standard Time)

Managed to fix it by changing these lines in the video_demo.py file:

checkpoint = torch.load(args['pre'], map_location='cuda:0')['state_dict']

checkpoint = torch.load(args['pre'], map_location='cuda:0')

Basically I added the map_location parameter and set it to my gpu ID