DSRL dual super-resolution learning for semantic segmentation We release our code in: https://github.com/Xilinx/DSRL