Dynamic EventNeRF
收藏资源简介:
Dynamic EventNeRF数据集由马克斯·普朗克信息学研究所和萨尔兰大学创建,旨在通过多视角事件流和稀疏RGB帧重建动态场景。该数据集包含多个场景和复杂的运动,涵盖从昏暗到非常黑暗的光照条件。数据集的创建过程涉及使用六台静态事件相机组成的真实世界多视角相机装置进行录制,并通过时间条件化的NeRF模型进行训练。该数据集主要应用于计算机视觉领域,特别是动态场景的4D重建,旨在解决在快速运动和低光照条件下传统RGB相机难以捕捉的问题。
The Dynamic EventNeRF dataset was developed by the Max Planck Institute for Informatics and Saarland University, with the goal of reconstructing dynamic scenes using multi-view event streams and sparse RGB frames. This dataset encompasses multiple scenes with complex motions, covering lighting conditions ranging from dim to extremely dark environments. The dataset creation process involved recording data via a real-world multi-view camera setup comprising six static event cameras, and training with time-conditioned NeRF models. Primarily utilized in the field of computer vision—especially for 4D reconstruction of dynamic scenes—this dataset aims to address the limitations of traditional RGB cameras when capturing fast-moving scenes under low-light conditions.




