PointInfinity：解析度不變的點擴散模型

PointInfinity: Resolution-Invariant Point Diffusion Models

April 4, 2024

作者: Zixuan Huang, Justin Johnson, Shoubhik Debnath, James M. Rehg, Chao-Yuan Wu

cs.AI

摘要

我們提出了PointInfinity，一個高效的點雲擴散模型系列。我們的核心思想是使用基於Transformer的架構，具有固定大小、與解析度無關的潛在表示。這使得在低解析度點雲上進行高效訓練成為可能，同時允許在推論期間生成高解析度的點雲。更重要的是，我們展示了在訓練解析度之外擴展測試解析度可以提高生成的點雲和表面的保真度。我們分析了這一現象，並將其與擴散模型中常用的無分類器引導相關聯，表明兩者都允許在推論期間在保真度和變異性之間進行折衷。在CO3D上的實驗表明，PointInfinity能夠高效生成高解析度的點雲（高達131k個點，比Point-E多31倍），並具有最先進的質量。

English

We present PointInfinity, an efficient family of point cloud diffusion models. Our core idea is to use a transformer-based architecture with a fixed-size, resolution-invariant latent representation. This enables efficient training with low-resolution point clouds, while allowing high-resolution point clouds to be generated during inference. More importantly, we show that scaling the test-time resolution beyond the training resolution improves the fidelity of generated point clouds and surfaces. We analyze this phenomenon and draw a link to classifier-free guidance commonly used in diffusion models, demonstrating that both allow trading off fidelity and variability during inference. Experiments on CO3D show that PointInfinity can efficiently generate high-resolution point clouds (up to 131k points, 31 times more than Point-E) with state-of-the-art quality.