Datasets for 3D Reconstruction & 3D Deep Learning
I collect a set of datasets for the sake of 3D reconstruction, hyperlinks for downloading are given below:
1. Datasets for 3D Reconstruction (SfM & MVS)
Internet Images
Small Object
Outdoor Scenes
Dataset | Hosted Source | Images | Size | Hyperlinks | Ground Truth |
---|---|---|---|---|---|
Pozzoveggiani | SAMANTHA | 54 | 4.7 MB | http://www.diegm.uniud.it/~fusiello/demo/samantha/ | Yes |
Piazza Dante | SAMANTHA | 39 | 5.3 MB | http://www.diegm.uniud.it/~fusiello/demo/samantha/ | Yes |
Piazza Erbe | SAMANTHA | 259 | 24 MB | http://www.diegm.uniud.it/~fusiello/demo/samantha/ | Yes |
Piazza Bra | SAMANTHA | 380 | 821 MB | http://www.diegm.uniud.it/~fusiello/demo/samantha/ | Yes |
Piazza Bra (NEW) | SAMANTHA | 331 | 358 MB | http://www.diegm.uniud.it/~fusiello/demo/samantha/ | Yes |
San Giacomo | SAMANTHA | 269 | 522 MB | http://www.diegm.uniud.it/~fusiello/demo/samantha/ | Yes |
Navona | SAMANTHA | 92 | 432 MB | http://www.diegm.uniud.it/~fusiello/demo/samantha/ | Yes |
gerrard-hall | colmap | 100 | 886.1 MB | https://drive.google.com/drive/folders/0B6q7-Pen0AbDTk5WM2hkUjF0Znc | Yes |
person-hall | colmap | 330 | 3.2 GB | https://drive.google.com/drive/folders/0B6q7-Pen0AbDTk5WM2hkUjF0Znc | Yes |
south-building | colmap | 128 | 233.9 MB | https://drive.google.com/drive/folders/0B6q7-Pen0AbDTk5WM2hkUjF0Znc | Yes |
Large-Scale Indoor Scenes
Dataset | Hosted Source | Images | Size | Hyperlinks | Ground Truth |
---|---|---|---|---|---|
TUM-LSI | TUM | 234 | - | https://github.com/NavVisResearch/NavVis-Indoor-Dataset/ | Yes |
Tanks & Temple | Tanks & Temple | - | - | https://www.tanksandtemples.org/download/ | Yes |
Google Streetview
Dataset | Hosted Source | Images | Size | Hyperlinks | Ground Truth |
---|---|---|---|---|---|
Streetview | INF | INF | https://console.cloud.google.com/storage/browser/streetview_image_pose_3d/?pli=1 | Yes |
Aerial Datasets
Dataset | Hosted Source | Images | Size | Hyperlinks | Ground Truth |
---|---|---|---|---|---|
PKU | PKU GIL Lab | - | - | - | waitting for uploading |
2. Datasets for Semantic Segmentation
Dataset | Hosted Source | Images | Size | Hyperlinks | Type |
---|---|---|---|---|---|
KITTI | TUM | - | - | http://www.cvlibs.net/datasets/kitti/eval_semantics.php | Dynamic Scenes |
PKU | PKU GIL Lab | - | - | https://github.com/PKUGIL/PSSI | Aerial Photo |
Cityscapes | Cityscapes | - | - | https://www.cityscapes-dataset.com/ | Urban Street Scenes |
3. Datasets for Feature&Descriptor Learning
waitting for updating
4. Datasets for Depth Estimation
waitting for updating
5. Datasets for Pose Estimation
Dataset | Hosted Source | Images | Size | Hyperlinks | Type | Ground Truth |
---|---|---|---|---|---|---|
Willow and Challenge Dataset | Willow Garage | - | 4.4GB | https://repo.acin.tuwien.ac.at/tmp/permanent/dataset_index.php | RGB-D frames | Yes |
TUW | Automation and Control Institute Tuwien | - | 2.43GB | https://repo.acin.tuwien.ac.at/tmp/permanent/dataset_index.php | RGB-D frames | Yes |
T-LESS | CVUT | - | - | http://cmp.felk.cvut.cz/t-less/ | RGB-D frames | Yes |
SXID | CVUT | 7datasets | - | http://cmp.felk.cvut.cz/sixd/challenge_2017/ | RGB-D frames | Yes |
ITODD | MVTec | - | 60GB | http://www.mvtec.com/company/research/datasets/mvtec-itodd/ | RGB-D frames | Yes |
致谢
感谢为此份数据集贡献自己力量的朋友们(排名不分先后,贡献不分大小):孙斌、陈煜