Unsupervised multimodal feature learning for semantic image segmentation | IEEE Conference Publication | IEEE Xplore