wengflow/robust-e-nerf
收藏资源简介:
Robust e-NeRF合成事件数据集用于研究移动事件相机在不同相机速度、对比度阈值变化和不定期周期下对NeRF重建质量的影响。该数据集通过改进的ESIM模拟器生成,包含三个不同难度级别的相机配置(简单、中等、困难),共21个序列记录。数据集结构按场景和难度级别组织,每个序列记录以ROS bag形式提供,包含事件、相机姿态、IMU测量、RGB图像、深度图、光流图和相机内参等数据流。此外,数据集还提供了验证和测试视图,适用于NeRF重建、新视角合成、3D重建、定位和SLAM等任务。
The Robust e-NeRF Synthetic Event Dataset is used to study the collective effect of camera speed profile, contrast threshold variation, and refractory period on the quality of NeRF reconstruction from a moving event camera. This dataset includes synthetic event data from three difficulty levels across seven Realistic Synthetic 360 scenes, totaling 21 sequence recordings. Each recording contains various data streams such as events, camera pose, IMU measurements, RGB images, depth maps, optical flow maps, and camera intrinsic parameters. The dataset is organized by scene and difficulty level, and includes validation and test views for each scene. It is designed for comparative analysis between event-based and image-based NeRF reconstruction methods and is suitable for various applications including novel view synthesis, 3D reconstruction, localization, and SLAM.
Robust e-NeRF Synthetic Event Dataset
概述
该数据集用于研究移动事件相机在不同相机速度配置、对比度阈值变化和不应期条件下的NeRF重建质量。数据集通过改进的ESIM模拟器生成,包含三种不同难度级别的相机配置(easy, medium, hard),共涉及七个真实感合成360场景,总计21个序列记录。
数据集结构和内容
数据集按场景和难度级别组织,每个序列记录以ROS bag形式存储,包含以下数据流:
| ROS Topic | Data | Publishing Rate (Hz) |
|---|---|---|
/cam0/events |
事件 | - |
/cam0/pose |
相机姿态 | 1000 |
/imu |
带模拟噪声的IMU测量 | 1000 |
/cam0/image_raw |
RGB图像 | 250 |
/cam0/depthmap |
深度图 | 10 |
/cam0/optic_flow |
光流图 | 10 |
/cam0/camera_info |
相机内参和镜头畸变参数 | 10 |
数据集通过运行改进的ESIM并使用关联的esim.conf配置文件生成,该文件引用相机内参配置文件pinhole_mono_nodistort_f={1111, 1250}.yaml和相机轨迹CSV文件{hemisphere, sphere}_spiral-rev=4[...].csv。
验证和测试视图
每个场景的验证和测试视图存储在views/文件夹中,结构与NeRF合成数据集一致(深度和法线图除外)。这些视图通过在Conda环境中安装Blender作为Python模块并运行bpy_render_views.py脚本生成。
引用
如果您使用此合成事件数据集进行研究,请引用以下论文:
bibtex @inproceedings{low2023_robust-e-nerf, title = {Robust e-NeRF: NeRF from Sparse & Noisy Events under Non-Uniform Motion}, author = {Low, Weng Fei and Lee, Gim Hee}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)}, year = {2023} }




