Building layout generation using site-embedded GAN model

Jiang, Feifeng; Ma, Jun; Webster, CJ; Li, Xiao; Gan, Vincent J.L.

doi:10.1016/j.autcon.2023.104888

Cited by 29 publications

(3 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We also investigate the influence of scene layouts, such as building contours and the spaces between buildings, on autonomous driving by editing or modifying the scene models. For example, we can quickly generate a large number of building layouts using the deep generative model ESGAN 9 . We simulate the perception of autonomous driving vehicles under different scene models to analyze the diverse perception results.…”

Section: Scene Modelsmentioning

confidence: 99%

See 1 more Smart Citation

Structured scene generation for autonomous driving simulation and digital twin scheduling

Li,

Ye,

Yang

et al. 2024

International Conference on Smart Transportation and City Engineering (STCE 2023)

View full text Add to dashboard Cite

Digital twin-based simulations play a crucial role in the development of autonomous driving and intelligent traffic systems. Autonomous driving simulation and digital twin scheduling involve the modeling, simulation, and control of the core components that impact the operation of autonomous driving. In this paper, we present a definition of the structured scenes, including scene models, topological road networks, high-definition maps, and driving patterns, that affect the autonomous driving process of vehicles. Previous studies primarily concentrated on the reconstruction or generation of scene models and topological road networks, while neglecting the application of autonomous driving. To alleviate this problem, we propose a structured scene generation framework tailored for autonomous driving. Besides, we conduct autonomous driving simulations and digital twin scheduling based on the structured scenes generated. In simulated tasks, the qualitative and quantitative results demonstrate that the group of autonomous vehicles exhibits efficient driving capabilities when operating under digital twin scheduling.

show abstract

Section: Scene Modelsmentioning

confidence: 99%

“…Ye et al 8 designed scene models in CARLA for downstream tasks. Jiang et al 9 proposed the site-embedded generative adversarial network (ESGAN) for automated building layout generation. It not only effectively generates 2D building layouts but also displays buildings in 3D with generated height information.…”

Section: Introductionmentioning

confidence: 99%

Structured scene generation for autonomous driving simulation and digital twin scheduling

Li,

Ye,

Yang

et al. 2024

International Conference on Smart Transportation and City Engineering (STCE 2023)

View full text Add to dashboard Cite

show abstract

“…In the field of automatic generation, computer-aided automatic spatial layout design methods have been proposed in the field of architecture for a long time [21][22][23], with the main goal of generating spatial layouts of buildings based on certain constraints. With the proposal of the Generative Adversarial Network (GAN) [24] and the research of related methods, more data-driven spatial layout design networks have been proposed, such as House-GAN [25], Building-GAN [26], House-GAN++ [27], and ESGAN [28], which utilize generative adversarial networks to extract regularities and automatically design spatial layouts based on constraint conditions.…”

Section: Introductionmentioning

confidence: 99%

A Generative Adversarial Network with Spatial Attention Mechanism for Building Structure Inference Based on Unmanned Aerial Vehicle Remote Sensing Images

Chen,

Guo,

Meng

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

The acquisition of building structures has broad applications across various fields. However, existing methods for inferring building structures predominantly depend on manual expertise, lacking sufficient automation. To tackle this challenge, we propose a building structure inference network that utilizes UAV remote sensing images, with the PIX2PIX network serving as the foundational framework. We enhance the generator by incorporating an additive attention module that performs multi-scale feature fusion, enabling the combination of features from diverse spatial resolutions of the feature map. This modification enhances the model’s capability to emphasize global relationships during the mapping process. To ensure the completeness of line elements in the generator’s output, we design a novel loss function based on the Hough transform. A line penalty term is introduced that transforms the output of the generator and ground truth to the Hough domain due to the original loss function’s inability to effectively constrain the completeness of straight-line elements in the generated results in the spatial domain. A dataset of the appearance features obtained from UAV remote sensing images and the internal floor plan structure is made. Using UAV remote sensing images of multi-story residential buildings, high-rise residential buildings, and office buildings as test collections, the experimental results show that our method has better performance in inferring a room’s layout and the locations of load-bearing columns, achieving an average improvement of 11.2% and 21.1% over PIX2PIX in terms of the IoU and RMSE, respectively.

show abstract

Revolutionizing wildlife protection: a novel approach combining deep learning and night-time surveillance

Madhasu,

Pande

2024

Multimed Tools Appl

View full text Add to dashboard Cite

Building layout generation using site-embedded GAN model

Cited by 29 publications

References 36 publications

Structured scene generation for autonomous driving simulation and digital twin scheduling

Structured scene generation for autonomous driving simulation and digital twin scheduling

A Generative Adversarial Network with Spatial Attention Mechanism for Building Structure Inference Based on Unmanned Aerial Vehicle Remote Sensing Images

Revolutionizing wildlife protection: a novel approach combining deep learning and night-time surveillance

Contact Info

Product

Resources

About