GUMBEL-NERF：将未见对象表示为部件组合式神经辐射场
GUMBEL-NERF: Representing Unseen Objects as Part-Compositional Neural Radiance Fields
AI总结：
针对 MoE NeRF 在少样本未见对象新视角合成中专家边界处的不连续问题，本文提出带事后专家选择机制的 Gumbel-NeRF，并在 SRN cars 上验证其图像质量优势。
摘要：
我们提出 Gumbel-NeRF，这是一种具有事后专家选择机制的混合专家（MoE）神经辐射场（NeRF）模型，用于合成未见对象的新视角。以往研究表明，MoE 结构能够对由许多对象组成的给定大规模场景提供高质量表示。然而，我们观察到，当把这种 MoE NeRF 模型用于从单样本/少样本输入合成未见对象的新视角任务时，它往往会在专家边界附近产生低质量表示。我们发现，这种退化主要由前瞻性专家选择机制造成，该机制可能在专家边界附近的对象形状中留下不自然的不连续性。Gumbel-NeRF 采用事后专家选择机制，即使在专家边界附近也能保证密度场的连续性。使用 SRN cars 数据集进行的实验表明，Gumbel-NeRF 在多种图像质量指标上均优于基线方法。
英文摘要：
We propose Gumbel-NeRF, a mixture-of-expert (MoE) neural radiance fields (NeRF) model with a hindsight expert selection mechanism for synthesizing novel views of unseen objects. Previous studies have shown that the MoE structure provides high-quality representations of a given large-scale scene consisting of many objects. However, we observe that such a MoE NeRF model often produces low-quality representations in the vicinity of experts' boundaries when applied to the task of novel view synthesis of an unseen object from one/few-shot input. We find that this deterioration is primarily caused by the foresight expert selection mechanism, which may leave an unnatural discontinuity in the object shape near the experts' boundaries. Gumbel-NeRF adopts a hindsight expert selection mechanism, which guarantees continuity in the density field even near the experts' boundaries. Experiments using the SRN cars dataset demonstrate the superiority of Gumbel-NeRF over the baselines in terms of various image quality metrics.
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。
