Example-based synthesis of 3D object arrangements

Fisher, Matthew; Ritchie, Daniel; Savva, Manolis; Funkhouser, Thomas; Hanrahan, Pat

doi:10.1145/2366145.2366154

Cited by 297 publications

(280 citation statements)

References 17 publications

Supporting

Mentioning

280

Contrasting

Order By: Relevance

“…To evaluate our framework, we tested our method on five datasets: the Princeton Shape Benchmark [28], the Stanford ShapeNetCore dataset [29], the Stanford Scene database [31], floor plans and food networks [32]. Table 1 shows the graph complexity of each dataset and the corresponding runtime performance.…”

Section: Resultsmentioning

confidence: 99%

Joint Graph Layouts for Visualizing Collections of Segmented Meshes

Ren

Schneider

Ovsjanikov

et al. 2018

IEEE Trans. Visual. Comput. Graphics

View full text Add to dashboard Cite

Section: Resultsmentioning

confidence: 99%

Joint Graph Layouts for Visualizing Collections of Segmented Meshes

Ren

Schneider

Ovsjanikov

et al. 2018

IEEE Trans. Visual. Comput. Graphics

View full text Add to dashboard Cite

“…Related questions include how to efficiently retrieve relevant 3D objects from a database such that they fit in a given scene context with high probability [Fisher and Hanrahan 2010;Fisher et al 2011]. New scenes can be synthesized after learning object relationships from a database of example scenes [Fisher et al 2012;Yu et al 2011;Yeh et al 2012]. The proposed mathematical models include probabilistic models based on Bayesian networks and Gaussian mixtures [Fisher et al 2012] and specialized probability distributions together with constraints encoded as factor graphs [Yeh et al 2012].…”

Section: Related Workmentioning

confidence: 99%

“…New scenes can be synthesized after learning object relationships from a database of example scenes [Fisher et al 2012;Yu et al 2011;Yeh et al 2012]. The proposed mathematical models include probabilistic models based on Bayesian networks and Gaussian mixtures [Fisher et al 2012] and specialized probability distributions together with constraints encoded as factor graphs [Yeh et al 2012]. The space spanned by these models is then appropriately sampled to derive plausible scenes.…”

Section: Related Workmentioning

confidence: 99%

“…While these methods have a broader scope than our work (e.g., they also need to pick the right type and number of objects), their placement models are limited compared with our approach. In Section 5, we compare our approach to one of the recent methods [Fisher et al 2012]. Additionally, our approach to provide examples at the level of individual operations has some advantages compared with using complete scenes as examples.…”

Section: Related Workmentioning

confidence: 99%

“…We additionally compare our work to the arrangement model used by Fisher et al [2012] by adapting our method to use a Gaussian kernel and the x-and y-position relative to an object centroid as geometric relationships. In Figure 11, two mirrors are placed on a bathroom wall so they have the same width as the sink and end at the ceiling.…”

Section: Computational Complexitymentioning

confidence: 99%

See 2 more Smart Citations

Learning shape placements by example

et al. 2015

View full text Add to dashboard Cite

KAUST propagate propagateFigure 1: New York-style skyscrapers generated by our method: Individual shape placement operations are learned during an example modeling session. Two of these operations are shown in the top row. New skyscraper variations like the ones in the bottom row can be generated from the learned model without additional user input. AbstractWe present a method to learn and propagate shape placements in 2D polygonal scenes from a few examples provided by a user. The placement of a shape is modeled as an oriented bounding box. Simple geometric relationships between this bounding box and nearby scene polygons define a feature set for the placement. The feature sets of all example placements are then used to learn a probabilistic model over all possible placements and scenes. With this model, we can generate a new set of placements with similar geometric relationships in any given scene. We introduce extensions that enable propagation and generation of shapes in 3D scenes, as well as the application of a learned modeling session to large scenes without additional user interaction. These concepts allow us to generate complex scenes with thousands of objects with relatively little user interaction.

show abstract

Building hierarchical structures for 3D scenes based on normalized cut

Zhao

Su²,

Yang³

2018

Computer Animation & Virtual

View full text Add to dashboard Cite

The growing number of 3D scene data available online brings in new challenges for scene retrieval, understanding, and synthesis. Traditional shape processing methods have difficulty to manage 3D scenes because such methods ignore the contextual information, that is, the spatial relationship between the objects or groups of objects, which plays a significant role in describing scenes. Therefore, a context-aware representation is needed to deal with such a problem. In this paper, we propose a method to build scene hierarchies based on contextual information. Given a 3D scene, we first use the interaction bisector surface to measure the affinity between different objects/elements of the scene and then apply the normalized cut method to build a hierarchical structure for the whole scene. The resulting hierarchical structure contains not only the relationship between the individual objects but also the relationship between object groups, which provides much richer information of the scene compared with a flat structure that only describes the contacts or affinity between the individual objects. We test our method using several public databases and show that the resulting structure is more consistent with the ground truth. We also show that our method can be used for point cloud segmentation and outperforms previous methods. KEYWORDShierarchical structure, normalized cut, scene analysis, segmentation INTRODUCTIONRecently, an increasing number of 3D scene databases have been available online because of the popularity of depth sensors and user-friendly modeling software. The ability to reuse such databases is critical for many content-hungry applications such as virtual worlds and games. To understand and retrieve existing 3D scene data, or synthesize new scenes based on existing ones, an essential problem is how to encode the scene information with a structural representation, which is compact, organized, and easy to use.Representing 3D scenes effectively is not a trivial task. The 3D scenes can be in a different format, such as depth data, point cloud data, 1 and artificial 3D models. 2 No matter what the format is, the representation should be able to encode not only the geometry of each object but also, more importantly, the contextual information of the scene. The spatial relationship between objects or object groups plays a significant role in describing the scene context. It describes how one object locates relative to another object, that is, A is on top of B, or A hooks on B, and how one object group locates relative to other groups, that is, group A is beside group B, or groups A and B are quite close while group C is a bit far away.

show abstract

Example-based synthesis of 3D object arrangements

Cited by 297 publications

References 17 publications

Joint Graph Layouts for Visualizing Collections of Segmented Meshes

Joint Graph Layouts for Visualizing Collections of Segmented Meshes

Learning shape placements by example

Building hierarchical structures for 3D scenes based on normalized cut

Contact Info

Product

Resources

About