How to control detection border thickness

[Arnesh1411]

Hi, I am working on instance segmentation of algae in microscopy images with a trained detectron2 model. After running the
model with get_sliced_prediction function, I have found the bbox's thickness so large that they hide the algae. Is there any way to control the border thickness? PFA

[fcakyon]

@Arnesh1411 you can set this parameter for custom box thickness size:

sahi/sahi/predict.py

Line 374 in 1240bab

visual_bbox_thickness: Optional[int] = None,

or set label text size with:

sahi/sahi/predict.py

Line 376 in 1240bab

visual_text_thickness: Optional[int] = None,

or hide labels with:

sahi/sahi/predict.py

Line 377 in 1240bab

visual_hide_labels: bool = False,