Is it possible to use the bounding box level augmentations for polygon datasets for instance segmentation?
Yes, it is possible to use bounding box level augmentations for polygon datasets for instance segmentation. Roboflow has introduced a new augmentation technique called bounding box level augmentation, which is not available anywhere else source.
However, if you have a dataset annotated with bounding boxes and want to convert them into instance segmentation annotation labels, you can use the SAM tutorial and notebook provided by Roboflow source.
For converting bounding boxes to polygons, you can use the
supervision.mask_to_xyxy utility or import detections into supervision and export from the mask property source.
Remember, polygons can always be converted back to bounding boxes, but bounding boxes cannot be as easily converted into polygons source.
If you already have a dataset annotated with bounding boxes and want to convert it into an instance segmentation dataset, you can use SAM standalone because the model can process existing bounding boxes source.
For labeling, Roboflow Annotate now offers automated polygon labeling for all users source.
Please note that the quality and quantity of the dataset have a significant impact on the performance of object detection models. High-quality annotations, with accurate object boundaries and precise labeling, play a crucial role in effectively training the model source.
Hi! I’m Roboflow’s AI powered bot. This is an automated response powered using AI with knowledge from our docs, blog, knowledge base, and previous forum responses. If you find this to be a helpful answer, please mark it as the solution using the checkbox icon below. If this doesn’t solve your issue, please follow up with what you’ve tried/why, and the community will continue to chime in as usual.
Here is the result after applying a 20% crop - bounding box level augmentation:
Why did the object become black?
Sorry to hear you’re having issues with generating your dataset. Could you please share either the Universe link or workspace/project ID so we can take a deeper look?
Here it is: Street Litter Instance Segmentation Dataset by Street Litter
I just generated the latest version and looked at the images.
There are still images of where the objects became black.
Hi! Any updates?
Sorry for the delayed response. Could you try regenerating the version? It’s possible this issue was fixed.