adding class names to the model checkpoint during training

[farukalamai]

I was checking the fintuning notebook of rf-detr. And I think it will be better if we directly access the class name from the model file for custom model.

Current approach:

import supervision as sv

ds = sv.DetectionDataset.from_coco(
    images_directory_path=f"{dataset.location}/test",
    annotations_path=f"{dataset.location}/test/_annotations.coco.json",
)

I have tried to change the main.py to save the class names in model checkpoint. But it's returning empty list. Do you have any suggestion what things I have to change? or maybe you will add that feature.

[SkalskiP]

@probicheaux have you been able to solve this with #105 ?

[farukalamai]

Hey @SkalskiP, I have opened a pr for this feature #119