Beyond Point Clouds: Scene Understanding by Reasoning Geometry and Physics

By Bo Zheng, Yibiao Zhao, Joey C. Yu, Katsushi Ikeuchi, Song-Chun Zhu. 2013

133 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computational Mechanics, Geology

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 Computational Mechanics

In Geology

Citations in the corpus

OctNetFusion: Learning Depth Fusion from Data (2017, background): “Shape completion from a single RGB-D image has been tackled in [14, 23, 45, 56].”
Semantic Scene Completion from a Single Depth Image (2016, background): “[37] which both predict binary voxel occupancy based on a single depth map without semantic understanding of the scene.”
"What Happens If..." Learning to Predict the Effect of Forces in Images (2016, background): “[43] reasons about the stability of objects in a given point cloud.”
Don't just listen, use your imagination: Leveraging visual common sense for non-visual tasks (2015, background): “common sense learnt from non-visual sources) benefits physical reasoning [23, 49], reasoning about intentions [40] and object functionality [50].”
To Fall Or Not To Fall: A Visual Approach to Physical Stability Prediction (2016, positive): “By including the additional clue from physical constraints into the inference mechanism, mostly from the support event, it has further improved results in segmentation of surfaces [12], scenes [26] from image data, and object segmentation in 3D point cloud data [31].”
Semantically-enriched 3D models for common-sense knowledge (2015, background): “The impact of gravity can be felt continuously by people and influences the structure of objects in the world.”
Structured Prediction of Unobserved Voxels from a Single Depth Image (2016, methodology): “It is not guaranteed that our results are physically stable, and how to best incorporate physics-based reasoning [61, 46] is still an open problem, but enforcing this prior may improve accuracy.”
Leveraging Visual Question Answering for Image-Caption Ranking (2016, background): “Recently there has been a surge of interest in visual reasoning tasks that require high-level reasoning such as physical reason-4 ing [34,35], future prediction [36,37,38], object affordance prediction [39] and textual tasks that require visual knowledge [40,41,42].”
Joint 3D Object and Layout Inference from a Single RGB-D Image (2015, background): “While the problem has traditionally been approached using RGB images [1, 8, 23, 36, 46] and videos [42], the availability of RGB-D sensors [30] and datasets [38] nourish the hope for more accurate models of the scene [11, 16, 18, 47].”
Anisotropic Convolutional Networks for 3D Semantic Scene Completion (2020, methodology): “The second dataset is the NYUCAD dataset [5].”
Dark, Beyond Deep: A Paradigm Shift to Cognitive AI with Humanlike Common Sense (2020, background): “jects should be physically stable and safe with respect to gravity and various other potential disturbances [97, 81, 98], such as an earthquake, a gust of wind, or the actions of other humans.”
Visual Stability Prediction and Its Application to Manipulation (2016, unknown): “By including additional clues from physical constraints into the inference mechanism, mostly from the support event, it has further improved results in segmentation of surfaces [17], scenes [18] from image data, and object segmentation in 3D point cloud data [19].”
Ieee Transactions on Pattern Analysis and Machine Intelligence 1 3d Reasoning from Blocks to Stability (background): “In concurrent work, Zheng et al. [39] reason about stability in a depth image.”
Scene Understanding by Reasoning Stability and Safety (2015, methodology): “In our recent work (Zheng et al. 2013, 2014), we propose a novel intuitive physics model based on gravity potential energy transfer.”
Sliding Shapes for 3D Object Detection in Depth Images (2014, methodology): “Semantic segmentation: A popular way of formulating object recognition in 3D is to predict the semantic label for each region of a depth map or 3D mesh [37, 38, 11, 10, 39–44].”
Configurable 3D Scene Synthesis and 2D Image Rendering with Per-pixel Ground Truth Using Stochastic Grammars (2017, methodology): “…2011; Zhao et al. 2013; Gupta et al. 2011; Jiang et al. 2013; Zhu et al. 2014; Myers et al. 2014; Koppula and Saxena 2014; Yu et al. 2015; Koppula and Saxena 2016; Roy and Todor-ovic 2016), and physical reasoning (Zheng et al. 2013, 2015; Zhu et al. 2015; Wu et al. 2015; Zhu et al. 2016; Wu 2016).”
Holistic++ Scene Understanding: Single-View 3D Holistic Scene Parsing and Human Pose Estimation With Human-Object Interaction and Physical Commonsense (2019, background): “By exploiting the underlying physical properties of scenes and objects, recent efforts have demonstrated the capability of estimating both current and future dynamics of static scenes [49, 28] and objects [60], understanding the support relationships and stability of objects [56], volumetric and occ”
Beat the MTurkers: Automatic Image Labeling from Weak 3D Supervision (2014, background): “In [38], the physical stability of objects is employed when segmenting point clouds.”