Fig. 4
From: Weakly supervised learning through box annotations for pig instance segmentation
The visualizations of instance segmentation results. The images in left column indicate ground truth annotation while the predicted masks and bounding boxes are shown in the right side.
