I am working on a project using Mask r-cnn to detect an object that is of a region of interest (Roi). When we do the training, the mask annotates using points (
Mask r-cnn