Open Access
Review
Table 4
Common datasets.
| Dataset | Release time | Data source | Number of models | Label category |
|---|---|---|---|---|
| KITTI | 2012 | Camera, LiDAR, GPS | 389 stereo images and optical flow pairs, labels for over 200k 3D targets | 2D/3D |
| DTU | 2016 | Camera, structured light scanner | 80 scenes with a total of 3920 views | 3D |
| ShapeNet | 2015 | CAD synthesis | Over 3 million models | 3D |
| ScanNet | 2017 | RGB-D camera | 1513 scenes with a total of 2.5 million images | 3D |
| Tanks and Temples | 2017 | Cameras, industrial laser scanners | 14 scenes, including objects such as tanks and artillery | 3D |
| ETH3D | 2017 | Professional DSLR camera, multi camera shooting platform, laser scanner | 13 training sets and 12 testing scenarios, 5 training and 5 testing videos, 27 training and 20 testing frames | 3D |
| ApolloScape | 2018 | Camera, LiDAR | 110K frames | 2D/3D |
| Semantic KITTI | 2019 | LiDAR, GPS | 28 labeling categories | 3D |
| BlendedMVS | 2020 | Camera | 113 3D models, totaling 17,818 images | 3D |
| BDD100K | 2020 | Camera, GPS | 100,000 HD videos | 2D |
| nuScenes | 2020 | Cameras, Radar, LiDAR | 1000 scenes tagged with 23 object categories totaling 14 million 3D tag boxes | 2D/3D |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.