SparseSplat: 向可应用的前馈3D高斯点划法迈进：像素不齐预测
SparseSplat: Towards Applicable Feed-Forward 3D Gaussian Splatting with Pixel-Unaligned Prediction
- Fudan University（复旦大学）
- ShanghaiTech University（上海科技大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文提出SparseSplat，首个可适应场景结构和局部区域信息丰富度的前馈3D高斯点划法模型，生成紧凑的3DGS地图，实验表明其在22%和1.5%的高斯数量下均能获得优异渲染质量。
AI中文摘要：
近期前馈3D高斯点划法（3DGS）的进步显著提升了渲染质量。然而，先前前馈3DGS方法生成的空间均匀且高度冗余的3DGS地图限制了其在下游重建任务中的整合。我们提出SparseSplat，首个前馈3DGS模型，能够根据场景结构和局部区域的信息丰富度自适应调整高斯密度，生成高度紧凑的3DGS地图。为此，我们提出基于熵的概率采样方法，生成纹理区域的大稀疏高斯和分配给信息丰富的区域的小密集高斯。此外，我们设计了专用点云网络，高效编码局部上下文并解码为3DGS属性，解决通用3DGS优化流程与前馈模型之间的感受野不匹配问题。广泛的实验结果表明，SparseSplat仅使用22%的高斯数量即可实现最先进的渲染质量，并且仅使用1.5%的高斯数量即可维持合理的渲染质量。项目页面：https://victkk.github.io/SparseSplat-page/.
英文摘要：
Recent progress in feed-forward 3D Gaussian Splatting (3DGS) has notably improved rendering quality. However, the spatially uniform and highly redundant 3DGS map generated by previous feed-forward 3DGS methods limits their integration into downstream reconstruction tasks. We propose SparseSplat, the first feed-forward 3DGS model that adaptively adjusts Gaussian density according to scene structure and information richness of local regions, yielding highly compact 3DGS maps. To achieve this, we propose entropy-based probabilistic sampling, generating large, sparse Gaussians in textureless areas and assigning small, dense Gaussians to regions with rich information. Additionally, we designed a specialized point cloud network that efficiently encodes local context and decodes it into 3DGS attributes, addressing the receptive field mismatch between the general 3DGS optimization pipeline and feed-forward models. Extensive experimental results demonstrate that SparseSplat can achieve state-of-the-art rendering quality with only 22% of the Gaussians and maintain reasonable rendering quality with only 1.5% of the Gaussians. Project page: https://victkk.github.io/SparseSplat-page/.
