scannet_temp
收藏资源简介:
本数据集是ScanNet数据集的重新打包版本,提供了完整的本地ScanNet数据集树的分割压缩存档。核心内容包括scans/目录(包含解压后可直接使用的多个3D室内场景扫描数据)和splits/目录(包含用于划分训练集、验证集和测试集的标准分割文件)。此版本排除了原始的场景级2D压缩文件、results/目录、color_90/目录以及其他特定文件(如pose_90.txt和selected_ids_90.txt)。数据以16个分卷的tar.gz文件(scannet_scans_part_000.tar.gz至scannet_scans_part_015.tar.gz)和一个单独的分割文件压缩包(scannet_splits.tar.gz)的形式提供,便于分步下载和管理。ScanNet数据集是一个大规模、丰富标注的3D室内场景数据集,广泛用于3D场景理解任务,如3D语义分割、实例分割和目标检测。
This dataset is a repackaged version of the ScanNet dataset, providing segmented compressed archives of the complete local ScanNet dataset directory tree. Its core content consists of the scans/ directory, which contains multiple ready-to-use decompressed 3D indoor scene scans, and the splits/ directory, which holds standard split files for partitioning the training, validation, and test sets. This release excludes the original scene-level 2D compressed files, the results/ directory, the color_90/ directory, and other specific files such as pose_90.txt and selected_ids_90.txt. The dataset is distributed as 16 split tar.gz archives (scannet_scans_part_000.tar.gz through scannet_scans_part_015.tar.gz) and a standalone compressed archive of split files (scannet_splits.tar.gz), enabling convenient step-by-step downloading and management. The ScanNet dataset is a large-scale, richly annotated 3D indoor scene dataset that is widely adopted for 3D scene understanding tasks including 3D semantic segmentation, instance segmentation, and object detection.
数据集概述:scannet_temp
该数据集是 ScanNet 完整本地数据集树的分割压缩归档版本,主要提供场景扫描数据与数据划分文件。
包含内容
- 场景数据:
scans/*目录下的场景,包含解压后可用的内容。 - 数据划分文件:
splits/*目录下的划分文件。
排除内容
- 场景级别的原始
*_2d-*.zip文件 results/目录color_90/目录pose_90.txt文件selected_ids_90.txt文件
文件列表
| 文件名 | 说明 |
|---|---|
scannet_scans_part_000.tar.gz ~ scannet_scans_part_015.tar.gz |
共16个场景数据分割压缩包(分片0-15) |
scannet_splits.tar.gz |
数据划分文件压缩包 |
下载与解压
-
下载命令: bash huggingface-cli download kairunwen/scannet_temp --repo-type dataset --local-dir ./scannet_temp_hf
-
解压命令: bash mkdir -p extracted for f in ./scannet_temp_hf/*.tar.gz; do tar -xzf "$f" -C extracted done




