Data annotation for mask rcnn

Is it mandatory to annotate images using polygon shapes for mask rcnn?
I read the and the research paper as well. It seems that matterport’s implementation can take bounding box as well as polygon as annotations. Although I am not certain. So should I consider bounding box annotation for my dataset? or polygon annotation?

Currently I have annotated some images using bounding box on Intel’s CVAT.

>Solution :

If you have a look COCO dataset, you can see it has 2 types of annotation format – bounding box and mask(polygon).
Therefore, Mast RCNN is to predict 3 outputs – Label prediction, Bounding box prediction, Mask prediction.
So, if you want Semantic Segmentation, you should have the polygon annotations for your dataset, but if you want only object detection, bounding box annotations are enough.

Leave a Reply