Learning to See the Invisible: End-to-End Trainable Amodal Instance Segmentation | IEEE Conference Publication | IEEE Xplore