Published February 6, 2024 | Version v1
Dataset Open

Data for Paper "Scalable Semantic 3D Mapping of Coral Reefs with Deep Learning"

  • 1. ROR icon École Polytechnique Fédérale de Lausanne
  • 2. University of Lausanne

Description

Example Data for DeepReefMap

This dataset contains input videos in MP4 format taken with GoPro Hero 10 Cameras in Reefs in the Red Sea to demonstrate the DeepReefMap tool, which is described in the paper "Scalable Semantic 3D Mapping of Coral Reefs with Deep Learning" by Sauder et al.

It contains a directory for model checkpoints for semantic segmentation, and for the 3D SLAM component:

```
checkpoints/
        segmentation_net.pth
        sfm_net.pth
```

It also contains videos to run the reconstruction with. See the detailed instructions for running reconstructions in https://github.com/josauder/mee-deepreefmap

```
input_videos/
        GX_SINGLE_VIDEO.MP4
        GX_VIDEO_1_OF_2.MP4
        GX_VIDEO_2_OF_2.MP4
```

Files

example_data.zip

Files (10.3 GB)

Name Size Download all
md5:b86bd6f8dc32df70150ea870ace81031
10.3 GB Preview Download
md5:93af84e3e4e6b2078356af9231d1a369
699 Bytes Preview Download

Additional details

Related works

Is part of
Software: 10.5281/zenodo.10649620 (DOI)
Is supplement to
Publication: 10.48550/arXiv.2309.12804 (DOI)
Publication: 10.1111/2041-210X.14307 (DOI)

Funding

Swiss National Science Foundation
- 205321 212614

Dates

Available
2024-02-06

Software

Repository URL
https://github.com/josauder/mee-deepreefmap
Programming language
Python
Development Status
Active