论文标题
从单视图中检测语义房的线框检测
Semantic Room Wireframe Detection from a Single View
论文作者
论文摘要
具有有限纹理信息或重复纹理的室内表面的重建，墙壁和天花板中常见的情况可能很困难，对于运动系统的单眼结构可能很困难。我们提出了一个语义室线框检测任务，以从单个角度图像预测语义线框。此类预测可以与形状先验一起使用，以估计房间布局和辅助重建。为了训练和测试所提出的算法，我们从模拟的结构3D数据集创建了一组新的注释。我们定性地表明，SRW-NET可以比以前的房间布局估计算法更好地处理复杂房间的几何形状，同时在非语义线框检测中定量地表现出基线。
Reconstruction of indoor surfaces with limited texture information or with repeated textures, a situation common in walls and ceilings, may be difficult with a monocular Structure from Motion system. We propose a Semantic Room Wireframe Detection task to predict a Semantic Wireframe from a single perspective image. Such predictions may be used with shape priors to estimate the Room Layout and aid reconstruction. To train and test the proposed algorithm we create a new set of annotations from the simulated Structured3D dataset. We show qualitatively that the SRW-Net handles complex room geometries better than previous Room Layout Estimation algorithms while quantitatively out-performing the baseline in non-semantic Wireframe Detection.
