how to keep only one class per object (i.e. local NMS per object)

[vhuard]

Search before asking

• I have searched the Roboflow Notebooks issues and found no similar bug report.

Description

i am using your notebook Automated Dataset Generation with Grounding DINO + Segment Anything Model (SAM)
on google colab
on some simple example from street views, Grounding Dino labels a car both as a "car" and as a "truck" though truck as a lower threshold.
i can play on one image on the box-threshold but i can't if i want to use the approach to do auto labelling on >100s of images.
how can i keep only the max threshold for same bounding boxes?

Additional

No response

Are you willing to submit a PR?

• Yes I'd like to help by submitting a PR!

[github-actions[bot]]

👋 Hello @vhuard, thank you for leaving an issue on Roboflow Notebooks.

🐞 Bug reports

If you are filing a bug report, please be as detailed as possible. This will help us more easily diagnose and resolve the problem you are facing. To learn more about contributing, check out our Contributing Guidelines.

If you require support with custom code that is not part of Roboflow Notebooks, please reach out on the Roboflow Forum or on the GitHub Discussions page associated with this repository.

💬 Get in touch

Do you have more questions about Roboflow that we haven't responded to yet? Feel free to ask them on the Roboflow Discuss forum. Our developer advocates and community team actively respond to questions there.

To ask questions about Notebooks, head over to the GitHub Discussions section of this repository.

[vhuard]

i haven't add custom code.
so just brought in my own pictures

[vhuard]

Can you help me on that?
I am still stuck to keep only the class with higher confidence when groundingdino is providing two identical boxes for two different classes

[SkalskiP]

Hi @vhuard 👋🏻! Sorry, I have been a bit busy lately. Let me convert this issue into a discussion thread and mode it to the Q&A section.

[SkalskiP]

Hi @vhuard 👋🏻! Have you tried using .with_nms? https://roboflow.github.io/supervision/detection/core/#supervision.detection.core.Detections.with_nms

[Syzygianinfern0]

@vhuard did you figure this out? I am meeting a similar issue.

[GeorgePearse]

yeah it will be

.with_nms(class_agnostic=True) 

The problem I have is that it's very slow because it's numpy not torch. But still a much neater implementation than I've seen anywhere else.