NVlabs
How can we use DOPE without camera
Hi, I'm using wsl2 ubuntu 20.04 and I've tried using usbipd to connect to a usb camera but it doesn't work, so I'm wondering if it's possible to use DOPE without a camera node? How do I do this, thanks!
You could use DOPE with images. Did you try the folder no_ros_inference?
https://github.com/NVlabs/Deep_Object_Pose/tree/master/scripts/train2
Sorry it was as the internal name. Check this part.
On Mon, Mar 6, 2023 at 15:39 Yukizyh @.***> wrote:
Hi, But I didn't find the no_ros_inference folder.
— Reply to this email directly, view it on GitHub https://github.com/NVlabs/Deep_Object_Pose/issues/286#issuecomment-1457220736, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABK6JIC64ZNIORJSHZG4MKLW2ZYUDANCNFSM6AAAAAAVQ4RQOY . You are receiving this because you commented.Message ID: @.***>
Thanks for replying. I tried the inference.py but there is some error:
I have tried changing the path to weight and using other objects like 'soup', but they all failed for the same reason : ERROR: Unable to find model weights: 'weights_dope/resnet_simple/alphabe_soup.pth'. I also tried to use the whole path of the weight, it doesn't work. And there is a warning here that the camera cannot be opened. I would like to know what are the changes that need to be made.
I created the path that matches the weight path in config file. So, it works now. Thanks a lot.
But there is an error that shows up when I try to generate the dataset:
And my cuda is 12.1. Does this problem cause by the wrong version cuda?
I will refer you on how to install nvisii, you need an NVIDIA GPU with up to date drivers.