Exploiting Multimodal Data In Computer Vision With Applications To Super-Resolution And Classification