We have a trained instance segmentation that we’re using via hosted inference, using workflows. In workflows, you can set IoU, but it’s not clear how it works. The documentation points to this: How to Code Non-Maximum Suppression (NMS) in Plain NumPy seems to refer to object detection specifically. …

This is the screenshot in the workflow mode: [Screenshot 2025-02-18 at 16.47.18] and the link goes here again - and the workflows documentation also takes you there.

Hi Reiner, Yes it looks like the NMS is only using axis aligned bounding boxes as far as I can tell. I think this is the code that does NMS post processing on instance segmentation in inference here: inference/inference/core/models/instance_segmentation_base.py at main · roboflow/inference · GitH…

Hi Thomas, thanks for the response, definitely helpful. I guess we can set the value to 0.99 (1 seems to error for some reason) and then do the filtering on our end afterwards. We’ll take a look at doing a PR too.

Does Intersection over Union filter out good detections when using instance segmentation?

🤝 Community Help

Reinier_Zeldenrust February 18, 2025, 4:49pm 2

This is the screenshot in the workflow mode:

and the link goes here again - and the workflows documentation also takes you there.

Topic		Replies	Views
Why is "overlap" not an option for instance segmentation models? 🤝 Community Help	3	381	August 7, 2024
How to use Double Detection Filter in Supervision library? 🤝 Community Help	4	245	June 18, 2024
Does an instance segmentation model benefit from both bounding box and polygon annotation data? 🤝 Community Help	2	252	July 14, 2023
Instance segmentation in roboflow 🤝 Community Help	1	63	June 28, 2024
Convert a detection project to an instance segmentation project 🤝 Community Help	1	302	October 17, 2023

Does Intersection over Union filter out good detections when using instance segmentation?

Related topics