README.md

metadata

license: cc-by-nc-4.0
tags:
  - 3d-reconstruction
  - photogrammetry
task_categories:
  - image-to-3d

EAH Dataset

Using a DJI Mavic Pro 2 drone, multiple videos were recorded of the three buildings at EAH Jena. Frames from these videos were extracted and post-processed to remove all visible people. These images were then used in Agisoft Metashape to generate 3D reconstructions of the buildings. The reconstructions were aligned to match the real-world orientation of the structures. Additionally, all intrinsic and extrinsic properties of each image were adjusted to accurately fit the resulting 3D models.

Details

• Photogrammetric 3D reconstructions of the three buildings of the EAH Jena.
• Each building comes with a set of images of size 3840 x 2160 px.
• Availability of intrinsic and extrinsic properties for each image.
• The capture time for each building is included.
• Published in the paper Enhanced Illumination Adjustment in 3D Outdoor Reconstructions via Shadow Removal through Color Transfer.

Figure 1: Visualization of all three building reconstructions, including rendered camera positions and example comparisons between real images and their corresponding rendered views.

Example: camera.json

This file contains the orientation for each camera:

[
    {
        "name": "2023-12-16_H123_S00_frame_000000",
        "matrix": [
            [
                -0.36218704547169456,
                0.0470962303839341,
                -0.02439450093992581,
                4.864710171746611
            ],
            [
                -0.011334934698936936,
                -0.2331992996061323,
                -0.28192589171372706,
                7.814374531518212
            ],
            [
                -0.05181376118229091,
                -0.2781953167233997,
                0.2321966890489866,
                -15.199963535169367
            ],
            [
                0.0,
                0.0,
                0.0,
                1.0
            ]
        ]
    },
    ...

Example: meta.json

This file contains the capture time range (start and end timestamps), the mesh center position (latitude and longitude) for each building, and the corresponding camera properties.

{
    "building": "123",
    "time": {
        "start": "2023-12-16T14:04:30Z",
        "end": "2023-12-16T14:47:00Z"
    },
    "position": {
        "latitude": 50.918489,
        "longitude": 11.568956
    },
    "camera" : {
        "width": 3840,
        "height": 2160,
        "f": 3099.1976427399018,
        "cx":-7.400794341491066,
        "cy":-25.280509058365258,
        "k1":-0.0027277484561097185,
        "k2":0.013177854331186076,
        "k3":-0.010827747284070439,
        "p1":-0.0015060061379312184,
        "p2":-0.0010258224524181574
    }
}

Citation

If you utilize this dataset in your research, kindly provide a citation:

@inproceedings{potechius2024,
    author = {Potechius, Herbert and Essaky, Selvam and Raja, Gunasekaran and Sikora, Thomas and Knorr, Sebastian},
    title = {Enhanced Illumination Adjustment in 3D Outdoor Reconstructions via Shadow Removal through Color Transfer},
    year = {2024},
    isbn = {9798400712814},
    publisher = {Association for Computing Machinery},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3697294.3697308},
    doi = {10.1145/3697294.3697308},
    booktitle = {Proceedings of 21st ACM SIGGRAPH Conference on Visual Media Production},
    articleno = {2},
    numpages = {10},
    location = {London, United Kingdom},
    series = {CVMP '24}
}