Conv-Deconv Grid-Network for semantic segmentation. Using only the training set without extra coarse annotated data (only 2975 images). No pre-training ( ...
確定! 回上一頁