delinqu/EN-SLAM-Dataset

EN-SLAM数据集是首个事件-RGBD隐式神经SLAM框架，能够有效利用事件流和RGBD克服极端运动模糊和光照变化场景的挑战。该数据集包括两个部分：DEV-Indoors和DEV-Reals。DEV-Indoors是通过Blender和模拟器获得的，包含正常、运动模糊和暗场景，共提供9个子集，包含RGB图像、深度图、事件流、网格和轨迹。DEV-Reals是从真实场景中捕获的，提供8个在运动模糊和光照变化下的挑战性子集。

The EN-SLAM dataset is the first event-RGBD implicit neural SLAM framework that efficiently leverages event stream and RGBD to overcome challenges in extreme motion blur and lighting variation scenes. The dataset includes two parts: DEV-Indoors and DEV-Reals. DEV-Indoors is obtained through Blender and simulator, covering normal, motion blur, and dark scenes, providing 9 subsets with RGB images, depth maps, event streams, meshes, and trajectories. DEV-Reals is captured from real scenes, providing 8 challenging subsets under motion blur and lighting variation.

原始信息汇总

数据集概述

数据集特征

• 名称: image
• 数据类型: image

数据集分割

• 分割名称: train
• 示例数量: 102293
• 数据大小: 44572171246.76 字节

数据集大小

• 总大小: 44572171246.76 字节