PlanIT

Wang, Kai; Lin, Yuan; Weissmann, Ben; Savva, Manolis; Chang, Angel X.; Ritchie, Daniel

doi:10.1145/3306346.3322941

Cited by 122 publications

(20 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Scene graphs have been shown to be powerful to generate 2D images [14,15]. Recent works [3,16,17,18,19,20] have been exploring leveraging scene graphs to guide 3D scene generation. To name a few, Luo et al [17] learned to generate furniture layout in a room given a scene graph as input.…”

Section: Related Workmentioning

confidence: 99%

“…More recently, deep learning-based methods further boost performance. PlanIT [3] introduced an image-based generative model reasoning over relation graphs. Ritchie et al [39] and Wang et al [40] proposed to learn image-based deep convolutional generative models.…”

Section: Related Workmentioning

confidence: 99%

“…Figure 5 (b) illustrates an example of collinearly parallel objects. Different from superstructures (Hub and Spokes, Chains) in PlanIT [3], since our representation takes the part into consideration, the hyperedges are detected by calculating the object part bounding box instead of the bounding box of the whole shape. Further, the orientation of the object is involved in the detection: 1. the orientation of all objects that satisfy the Parallel Collinearity must be the same; 2. the orientation of all objects that satisfy the N -fold Rotational Symmetry must point to the same center object.…”

Section: They Satisfy a Collinearly Parallel Hyperedge If And Only Ifmentioning

confidence: 99%

“…[1,2]) and more recent deep generative models (e.g. [3,4]). Although such works can generate diverse and plausible furniture layouts, they do not generate furniture geometry at the same time, and usually only retrieve existing furniture shapes from a repository.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

SceneHGN: Hierarchical Graph Networks for 3D Indoor Scene Generation With Fine-Grained Geometry

Gao

Sun

et al. 2023

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

3D indoor scenes are widely used in computer graphics, with applications ranging from interior design to gaming to virtual and augmented reality. They also contain rich information, including room layout, as well as furniture type, geometry, and placement. High-quality 3D indoor scenes are highly demanded while it requires expertise and is time-consuming to design high-quality 3D indoor scenes manually. Existing research only addresses partial problems: some works learn to generate room layout, and other works focus on generating detailed structure and geometry of individual furniture objects. However, these partial steps are related and should be addressed together for optimal synthesis. We propose SCENEHGN, a hierarchical graph network for 3D indoor scenes that takes into account the full hierarchy from the room level to the object level, then finally to the object part level. Therefore for the first time, our method is able to directly generate plausible 3D room content, including furniture objects with fine-grained geometry, and their layout. To address the challenge, we introduce functional regions as intermediate proxies between the room and object levels to make learning more manageable. To ensure plausibility, our graph-based representation incorporates both vertical edges connecting child nodes with parent nodes from different levels, and horizontal edges encoding relationships between nodes at the same level. Our generation network is a conditional recursive neural network (RvNN) based variational autoencoder (VAE) that learns to generate detailed content with fine-grained geometry for a room, given the room boundary as the condition. Extensive experiments demonstrate that our method produces superior generation results, even when comparing results of partial steps with alternative methods that can only achieve these. We also demonstrate that our method is effective for various applications such as part-level room editing, room interpolation, and room generation by arbitrary room boundaries.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: They Satisfy a Collinearly Parallel Hyperedge If And Only Ifmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

SceneHGN: Hierarchical Graph Networks for 3D Indoor Scene Generation With Fine-Grained Geometry

Gao

Sun

et al. 2023

IEEE Trans. Pattern Anal. Mach. Intell.

View full text Add to dashboard Cite

show abstract

“…This paper focuses on converting an input graph to a realistic house footprint, as depicted in Figure 1. Existing house generation methods such as [2,15,18,26,30,43,45], typically rely on building convolutional layers. However, convolutional architectures lack an understanding of longrange dependencies in the input graph since inherent inductive biases exist.…”

Section: Introductionmentioning

confidence: 99%

Graph Transformer GANs for Graph-Constrained House Generation

Tang¹,

Zhang²,

Li³

et al. 2023

Preprint

View full text Add to dashboard Cite

We present a novel graph Transformer generative adversarial network (GTGAN) to learn effective graph node relations in an end-to-end fashion for the challenging graphconstrained house generation task. The proposed graph-Transformer-based generator includes a novel graph Transformer encoder that combines graph convolutions and selfattentions in a Transformer to model both local and global interactions across connected and non-connected graph nodes. Specifically, the proposed connected node attention (CNA) and non-connected node attention (NNA) aim to capture the global relations across connected nodes and non-connected nodes in the input graph, respectively. The proposed graph modeling block (GMB) aims to exploit local vertex interactions based on a house layout topology. Moreover, we propose a new node classification-based discriminator to preserve the high-level semantic and discriminative node features for different house components. Finally, we propose a novel graph-based cycle-consistency loss that aims at maintaining the relative spatial relationships between ground truth and predicted graphs. Experiments on two challenging graph-constrained house generation tasks (i.e., house layout and roof generation) with two public datasets demonstrate the effectiveness of GTGAN in terms of objective quantitative scores and subjective visual realism. New state-of-the-art results are established by large margins on both tasks.

show abstract