Exploring Video Frame Redundancies for Efficient Data Sampling and Annotation in Instance Segmentation