Add ultralytics SAM support when using bboxes and points arguments in SAM().predict()

[xaristeidou]

Description

Fixes issue noted in #1453 .

Added brief check and fill of necessary fields in sv.Detections()

List any dependencies that are required for this change.

No new dependencies

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

How has this change been tested, please provide a testcase or example of how you tested the change?

Colab attached for testing

Docs

• Docs updated? Not needed

Colab notebook for testing: https://colab.research.google.com/drive/10aRaBAfb__xrMeBNn9amYwrKgcnbExBn?usp=sharing

[xaristeidou]

class_id=np.arange(len(ultralytics_results)), : This is due to the fact that in ultralytics Results if there are more than one masks, their class_id is labeled with increasing +1 step start from 0 (e.g. 0,1,2, etc).

[xaristeidou]

Why does pre-commit check fails, mask_to_xyxy is imported properly and runs as expected locally. Locally in my system pre-commit pass all checks.

[onuralpszr]

@xaristeidou I also added mask to xyxy and pre-commit locally fine at my end. Probably some server issue. I tested with normal cases such as yolov8 model not with sam and works well. LGTM to me !

[onuralpszr]

Why does pre-commit check fails, mask_to_xyxy is imported properly and runs as expected locally. Locally in my system pre-commit pass all checks.

I think server issue I have no idea

[onuralpszr]

@xaristeidou I am going to update your branch to latest dev and test that out before merge maybe pre-commit will act fine I will backup and revert in case something happen.

[onuralpszr]

@xaristeidou I understand why act weird because in our CURRENT develop we removed that import so compare to latest makes fail so I had to update repo and re-add again to fix it. Now it is all good.

[xaristeidou]

@onuralpszr Nice! Makes sense now.

[onuralpszr]

I am merging in. Thank you so much @xaristeidou !