Unfolding an Indoor Origami World

By David F. Fouhey, Abhinav Kumar Gupta, Martial Hebert. 2014

103 citations recorded by the source corpus.

In this work, we present a method for single-view reasoning about 3D surfaces and their relationships. We propose the use of mid-level constraints for 3D scene understanding in the form of convex and concave edges and introduce a generic framework capable of incorporating these and other constraints. Our method takes a variety of cues and uses them to infer a consistent interpretation of the scene. We demonstrate improvements over the state-of-the art and produce interpretations of the scene that link large planar surfaces.

Subfields: Aerospace Engineering, Computer Graphics and Computer-Aided Design, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Graphics and Computer-Aided Design

In Computer Vision and Pattern Recognition

Citations in the corpus

Transitive Invariance for Self-Supervised Visual Representation Learning (2017, methodology): “We follow the same protocols for generating surface normal ground truth and evaluations as [14, 28, 15].”
Matterport3D: Learning from RGB-D Data in Indoor Environments (2017, methodology): “These previous RGB-D datasets have been used to train models for several standard scene understanding tasks, including semantic segmentation [27, 33, 13, 40, 7], 3D object detection [31, 24, 14, 36, 37], normal estimation [41, 23, 9, 3], camera relocalization [30, 39], and others [47, 11, 12].”
Marr Revisited: 2D-3D Alignment via Surface Normal Prediction (2016, background): “surface normal and depth map) by reasoning about mid-level scene properties, such as discriminative 3D primitives [10], convex and concave edges [11], and style elements harvested by unsupervised learning [12].”
Semi-global Stereo Matching with Surface Orientation Priors (2017, methodology): “, expected ground plane orientation in autonomous driving [28]), from semantic analysis [15, 16, 2], or from vanishing point analysis and Manhattan-world assumptions [41, 43, 17, 32, 36], which we also explore in this paper.”
Single Image 3D without a Single 3D Image (2015, methodology): “At test time, this local mapping is used on a single image to infer the 3D labels; in other works, it is again presumed that there is such regularity that one can impose even more top-down constraints, such as the Manhattan-world assumption [8], an indoor box model [20, 35], or others [27, 6, 5, 17,”
Designing deep networks for surface normal estimation (2014, methodology): “We adopt the protocols introduced in [9] and used by state-of-the-art methods on this task [9, 10, 22].”
Three-Dimensional Object Detection and Layout Prediction Using Clouds of Oriented Gradients (2016, background): “Recent advances in depth sensor technology have greatly reduced the ambiguities present in standard RGB images, enabling breakthroughs in scene layout prediction [22, 13, 41], support surface prediction [34, 8, 10], semantic parsing [11], and object detection [36].”
Enforcing Geometric Constraints of Virtual Normal for Depth Prediction (2019, background): “Several datadriven methods [8, 9, 11, 12, 45, 49] have achieved promising results.”
Unsupervised Learning of Visual Representations Using Videos (2015, methodology): “For evaluation and generating the groundtruth, we adopt the protocols introduced in [12] which is used by different methods [12, 25, 13] on this task.”
Depth and surface normal estimation from monocular images using regression on deep features and hierarchical CRFs (2015, methodology): “When specific scene dependant knowledge is available, depth estimation or 3D reconstruction from single images can be achieved by utilizing geometric assumptions such as the “Blocks World” model [5], the “Origami World” model [6], shape from shading [7] and repetition of structures [8].”
Just Look at the Image: Viewpoint-Specific Surface Normal Prediction for Improved Multi-View Reconstruction (2016, methodology): “Later that method was augmented with shape priors for rooms and an explicit model of crease edges [12].”
Pattern-Affinitive Propagation Across Depth, Surface Normal and Semantic Segmentation (2019, methodology): “Surface Normal Estimation: Recent methods designed for surface normal estimation are mainly based on deep neural networks [13, 14, 61, 55].”
GeoNet: Geometric Neural Network for Joint Depth and Surface Normal Estimation (2018, unknown): “8 UNFOLD [10] 35.”
Learning to See by Moving (2015, methodology): “The two streams of the SCNN share the same architecture and the same set of weights and consequently perform the same set of operations for computing features.”
Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-scale Convolutional Architecture (2014, background): “timization on a grid drawn from vanishing point rays [11], while Ladicky et al.”
SUN RGB-D: A RGB-D scene understanding benchmark suite (2015, background): “There are many interesting works on RGB-D scene understanding, including semantic segmentation [53, 49, 19] object classification [69], object detection [59, 20, 62], context reasoning [38], mid-level recognition [32, 31], and surface orientation and room layout estimation [13, 14, 74].”