Replies: 1 comment 5 replies
-
@baleris you can try multi_label, agnostic and iou_thresh settings during inference. |
Beta Was this translation helpful? Give feedback.
5 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
@glenn-jocher
i have a trained model on custom dataset which is detecting 5 classes from an image. But i see there were some bounding boxes detected inside another detected bounding box.
For example, model detects a class-1 object which is having a size of around 100x100(bbox) and i see another class-2 of size 10x20(bbox) detected inside the area of class-1 detected bounding box. Also i see class-1 may re appear inside the already detected area of class-1 i.e i may have another smaller object of same class can appear inside the larger class already detected.
I would like to consider only outer larger bounding box which is detected by model and want to avoid any classes detected inside that area. Is there any possibility in general.py or any other configuration i can modify to achieve this ?
I tried these agnostic & multi_label options(below), wouldn't worked for me.
#Update inference default to multi_label=False #2252
#Limit bounding boxes #2248
Beta Was this translation helpful? Give feedback.
All reactions