GAMES Webinar 2019 – 98期（SIGGRAPH 2019 报告） | 吴志杰（深圳大学），李曼祎（山东大学）
【GAMES Webinar 2019-98期】
报告题目：SAGNet: Structure-aware Generative Network for 3D-Shape Modeling
We present SAGNet, a structure-aware generative model for 3D shapes. Given a set of segmented objects of a certain class, the geometry of their parts and the pairwise relationships between them (the structure) are jointly learned and embedded in a latent space by an autoencoder. The encoder intertwines the geometry and structure features into a single latent code, while the decoder disentangles the features and reconstructs the geometry and structure of the 3D model. Our autoencoder consists of two branches, one for the structure and one for the geometry. The key idea is that during the analysis, the two branches exchange information between them, thereby learning the dependencies between structure and geometry and encoding two augmented features, which are then fused into a single latent code. This explicit intertwining of information enables separately controlling the geometry and the structure of the generated models. We evaluate the performance of our method and conduct an ablation study. We explicitly show that encoding of shapes accounts for both similarities in structure and geometry. A variety of quality results generated by SAGNet are presented.
Zhijie Wu, a graduate of Sichuan University, is currently working as an intern student under supervision of Prof. Hui Huang at the Visual Computing Research Center of Shenzhen University. His recent research interests are deep learning based computer graphics, including 3D generative model, image generation, style transfer and so on.
报告题目：GRAINS: Generative Recursive Autoencoders for INdoor Scenes
We present a generative neural network for generating plausible 3D indoor scenes in large quantities and varieties easily and highly efficiently. Observing that indoor scene structures are inherently hierarchical, a variational recursive autoencoder is trained to perform hierarchical scene object grouping during encoding phase and hierarchical scene generation during decoding. We demonstrate the capability of our method to generate plausible and diverse 3D indoor scenes and show the applications.
Manyi Li is a Ph.D from the School of Computer Science in Shandong University, supervised by Prof. Changhe Tu. From Sep. 2017 to Sep. 2018, she spent one year visiting the GrUVi Lab in the School of Computing Science at Simon Fraser University, Canada, under the supervision of Prof. Hao (Richard) Zhang.
Her research focus on Computer Graphics, especially on the analysis and synthesis of 3D objects and scenes.
GAMES主页的“使用教程”中有 “如何观看GAMES Webinar直播？”及“如何加入GAMES微信群？”的信息；