-
Notifications
You must be signed in to change notification settings - Fork 6.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
prioritize batching for torchvision::nms #8593
Comments
Hi @RishiMalhotra920 , can you please share a minimal reproducing example of the code you tried to run? Not that we do have a batched version of |
Ah unfortunately, I can't find the place where i saw the warning again but i think i was doing something with
Additionally, i saw the batched_nms earlier and this threw me off initially since it mentioned that it does not do NMS between objects in different categories. However, to make this work for multiple images in a batch, i would just assign each image in the batch with a different category and it would work as expected. Thanks for pointing me to this! Once you have noted the suggestion, if you don't have any other questions, you can close this. |
Thanks for the reply @RishiMalhotra920
Can you share a reproducible example of this? If all examples in a batch have a different label, then |
Thanks to the Team for developing PyTorch! It would be so much useful to have namely batch-NMS, but multiclass-NMS as now. It's extremely helpfull for for those who works with Triton and Jetson and needs to Please, it would be remarkable to have NMS which returns the bboxes with the batch dimension and padding with fake bboxes to support the tensor shape. |
🚀 The feature
Implement batched nms!
Motivation, pitch
My motivation to create this issue was this warning:
Alternatives
No response
Additional context
No response
The text was updated successfully, but these errors were encountered: