This novel, data-driven framework for semantic scene understanding works without pixelwise annotation or classifier training using a probabilistic Expectation-Maximization formulation. It performs better than state-of...
详细信息
This novel, data-driven framework for semantic scene understanding works without pixelwise annotation or classifier training using a probabilistic Expectation-Maximization formulation. It performs better than state-of-the-art methods in both semantic segmentation and image annotation.
暂无评论