arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.13408cs.CVcs.AIcs.LG
收藏

三维点云生成模型

Generative Models for 3D Point Clouds

Stanford University(斯坦福大学)

Lingjie Kong, Pankaj Rajak, Siamak Shakeri

摘要:

点云是一种丰富的几何数据结构,其三维结构为理解三维空间中的表示学习和生成式建模提供了一个很好的领域。在这项工作中,我们的目标是通过实验与Transformer编码器,潜在空间流模型,和自回归解码器,以提高点云潜在空间生成模型的性能。我们分析和比较这些模型的生成和重建性能的各种对象类型。

英文摘要:

Point clouds are rich geometric data structures, where their three dimensional structure offers an excellent domain for understanding the representation learning and generative modeling in 3D space. In this work, we aim to improve the performance of point cloud latent-space generative models by experimenting with transformer encoders, latent-space flow models, and autoregressive decoders. We analyze and compare both generation and reconstruction performance of these models on various object types.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑