Exploring generative 3D shapes using autoencoder networks

Umetani, Nobuyuki

doi:10.1145/3145749.3145758

Cited by 58 publications

(48 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They are commonly used for 3D reconstruction [41,18] but suffer from large-scale memory inefficiency and require a space discretization which loses fine-grained details. Recent works compress 3D shapes using auto-encoders to efficiently handle geometric information [58,56,12]. They typically encode-decode shapes into different representations.…”

Section: Related Workmentioning

confidence: 99%

Leveraging Shape Completion for 3D Siamese Tracking

Giancola

Zarzar

Ghanem

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

165

207

View full text Add to dashboard Cite

Point clouds are challenging to process due to their sparsity, therefore autonomous vehicles rely more on appearance attributes than pure geometric features. However, 3D LIDAR perception can provide crucial information for urban navigation in challenging light or weather conditions. In this paper, we investigate the versatility of Shape Completion for 3D Object Tracking in LIDAR point clouds. We design a Siamese tracker that encodes model and candidate shapes into a compact latent representation. We regularize the encoding by enforcing the latent representation to decode into an object model shape. We observe that 3D object tracking and 3D shape completion complement each other. Learning a more meaningful latent representation shows better discriminatory capabilities, leading to improved tracking performance. We test our method on the KITTI Tracking set using car 3D bounding boxes. Our model reaches a 76.94% Success rate and 81.38% Precision for 3D Object Tracking, with the shape completion regularization leading to an improvement of 3% in both metrics.

show abstract

Section: Related Workmentioning

confidence: 99%

Leveraging Shape Completion for 3D Siamese Tracking

Giancola

Zarzar

Ghanem

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

165

207

View full text Add to dashboard Cite

show abstract

“…In particular, variational autoencoder (VAE) and generative adversarial network (GAN) are popular generative models used in design optimization, where high-dimensional design variables are encoded in low-dimensional design space [13,14]. In addition, these models are utilized in the design exploration and shape parameterization [8,9].…”

Section: Generative Models For Generative Designmentioning

confidence: 99%

Deep Generative Design: Integration of Topology Optimization and Generative Models

Jung

Kim

et al. 2019

Journal of Mechanical Design

286

124

View full text Add to dashboard Cite

Deep learning has recently been applied to various research areas of design optimization. This study presents the need and effectiveness of adopting deep learning for generative design (or design exploration) research area. This work proposes an artificial intelligent (AI)-based deep generative design framework that is capable of generating numerous design options which are not only aesthetic but also optimized for engineering performance. The proposed framework integrates topology optimization and generative models (e.g., generative adversarial networks (GANs)) in an iterative manner to explore new design options, thus generating a large number of designs starting from limited previous design data. In addition, anomaly detection can evaluate the novelty of generated designs, thus helping designers choose among design options. The 2D wheel design problem is applied as a case study for validation of the proposed framework. The framework manifests better aesthetics, diversity, and robustness of generated designs than previous generative design methods. Nomenclature: expectation : differentiable function of discriminator : differentiable function of generator : noise variable ℒ: loss function of autoencoder c: compliance : density variable ̃: filtered density variable ̃̅ : projected density variable : penalization factor Generative Models for Generative DesignGenerative models, one of the promising deep learning areas, can enhance research on generative design. The generative model is an algorithm for constructing a generator that learns the probability distribution of training data and generates new data based on learned probability distribution. In particular, variational autoencoder (VAE) and generative adversarial network (GAN) are popular generative models used in design optimization, where high-dimensional design variables are encoded in low-dimensional design space [13,14]. In addition, these models are utilized in the design exploration and shape parameterization [8,9].The use of generative model to produce engineering designs directly is limited [23]. However, this study claims that the limitations can be overcome by integrating with topology optimization. First, the generative model requires a number of training data, but accumulated training data for various designs in the industry are confidential and difficult to access. A number of designs obtained from topology optimization are expected to serve as training data. Second, the generative model cannot guarantee feasible engineering. In this case,

show abstract

“…Overfitting can be a concern when the training dataset size is not large. While the size of our dataset is comparable to other training datasets described in learning literature [50,2], we would like our approach to be applicable to smaller or evolving datasets of character animation data.…”

Section: Overfittingmentioning

confidence: 99%