GL3D: Geometric Learning with 3D Reconstruction

About

GL3D (Geometric Learning with 3D Reconstruction) is a large-scale database created for 3D reconstruction and geometry-related learning problems. Most images contained are captured by drones from multiple scales and perspectives with large geometric overlaps, covering urban, rural area, or scenic spots. It also includes small object reconstructions to enrich the data diversity. If you find this dataset useful for your research, please cite:

@inproceedings{shen2018mirror,
    author={Shen, Tianwei and Luo, Zixin and Zhou, Lei and Zhang, Runze and Zhu, Siyu and Fang, Tian and Quan, Long},
    title={Matchable Image Retrieval by Learning from Surface Reconstruction},
    booktitle={The Asian Conference on Computer Vision (ACCV},
    year={2018},
}

If you have used the correspondence labels, please also cite:

@inproceedings{luo2018geodesc,
    title={Geodesc: Learning local descriptors by integrating geometry constraints},
    author={Luo, Zixin and Shen, Tianwei and Zhou, Lei and Zhu, Siyu and Zhang, Runze and Yao, Yao and Fang, Tian and Quan, Long},
    booktitle={European Conference on Computer Vision (ECCV)},
    year={2018}
}

Dataset Description

GL3D contains 90,630 high-resolution images regarding 378 different scenes. Each scene data is reconstructed to generate a triangular mesh model by the state-of-the-art 3D reconstruction pipeline. Refer to [1] for details. For each scene data, we provide the complete image sequence, geometric labels and reconstruction results.

Tasks

Research works below are supported by GL3D:

Task	Reference
Image retrieval	MIRorR, ACCV'18
Local descriptor	GeoDesc, ECCV'18
Local descriptor	ContextDesc, CVPR'19

Download

For image retrieval task, use 224x224 images and refer to MIRorR.

For learning local descriptor, use 1000x1000 images and refer to GeoDesc.

Sources	Data Name	Chunk Start	Chunk End	Descriptions
GL3D	gl3d_full_size	TBA	TBA	Full-size images of GL3D
GL3D	gl3d_224	0	6	224x224 images of GL3D
GL3D	gl3d_1000	0	91	1000x1000 images of GL3D

Use download_data.sh script to download the tar files, by passing augments

bash download_data.sh <data_name> <chunk_start> <chunk_end>

For example, to download GL3D 224x224 images, run

bash download_data.sh gl3d_224 0 6 

To extract the files, run

cat download_data_gl3d_224/*.tar.* | tar -xvf -

Dataset Format

data                          
 └── <pid> 
       ├── images/*
       ├── geolabel/*
       ├── img_kpts/*.bin
       └── image_list.txt

File Name	Data Name	Chunk Start	Chunk End	Task	Descriptions
geolabel/cameras.txt	gl3d_cams	0	0	Common	Camera intrisic/extrinsic parameters, recovered by SfM.
img_kpts/<img_idx>.bin	gl3d_kpts	0	58	Common	Image keypoints detected by SIFT.
geolabel/corr.bin	gl3d_corr	0	7	Local descriptor	Image correspondences that haved survived from SfM.
geolabel/mask.bin	gl3d_mask	0	8	Image retrieval	Overlap masks of image pairs, computed from mesh re-projections.
geolabel/mesh_overlap.txt	gl3d_mo	0	0	Image retrieval	Mesh overlap ratio of image pairs, computed from mesh re-projections.
geolabel/common_track.txt	gl3d_ct	0	0	Image retrieval	Common track ratio of image pairs, computed from SfM.

Again, use download_data.sh to fetch the above geometric labels or reconstruction results,

For data organization, refer to docs/format.md.

Python-based IO utilities are provided to parse the data, refer to utils/io.py.

Visualizations and examples of usage can be found in example/README.md.

Please feel free to inform us if you need some other intermediate results for your research.

Data Preview

The mesh reconstruction is available for preview by substituting <pid> in the following link:

https://www.altizure.com/project-model?pid=<pid>

An example is provided here. Noted that some projects are not online available, from 000000000000000000000000 to 00000000000000000000001d.

Acknowledgments

This dataset is prepared and maintained by Zixin Luo, Tianwei Shen, Jacky Tang and Tian Fang. 3D reconstructions are obtained by Altizure.