SpaRe
收藏资源简介:
SpaRe数据集是由华为诺亚方舟实验室开发,专门用于稀疏视角神经渲染的基准测试。该数据集包含102个精心挑选的合成场景,每个场景有高达64个相机视角和7种光照配置,分辨率为1600×1200。数据集的创建过程模拟了DTU数据集的设置,确保了相机姿态和光照位置的精确性。SpaRe数据集旨在解决稀疏视角下的神经渲染问题,适用于自动驾驶、增强现实和资源受限环境等实际应用场景。
The SpaRe dataset is developed by Huawei Noah's Ark Lab as a dedicated benchmark for sparse-view neural rendering. It comprises 102 carefully curated synthetic scenes, with each scene featuring up to 64 camera views and 7 lighting configurations at a resolution of 1600×1200. The dataset is constructed by simulating the experimental setup of the DTU dataset, ensuring the accuracy of camera poses and lighting positions. The SpaRe dataset is designed to address the neural rendering problem under sparse-view conditions, and is applicable to real-world scenarios such as autonomous driving, augmented reality, and resource-constrained environments.

- 1AIM 2024 Sparse Neural Rendering Challenge: Dataset and Benchmark华为诺亚方舟实验室 · 2024年



