Run with Public Datasets

SLAMcore software supports the following public datasets:

  • EuRoC MAV Dataset 1,

  • OpenLORIS-Scene Dataset 2, and

  • TUM VI Benchmark Dataset 3

To run SLAMcore with one of the public datasets:

  1. Download the sequences from the official websites,

  2. Install the slamcore_utils repository, and

  3. Run the slamcore-setup-dataset script, which embeds a configuration file that points to the correct camera calibration data and converts the public datasets to a format that SLAMcore supports.

  4. You should now be able to run the Visualiser using this dataset by running this in a terminal window:

    $ slamcore_visualiser dataset -u <path/to/sequence>
    

Note

For the EuRoC and TUM-VI datasets, please run SLAMcore on the dataset with mav0 as the root folder, for example:

$ slamcore_visualiser dataset -u /home/ubuntu/tum/dataset-room1_512_16/mav0/

Now you have confirmed that the software has downloaded and runs successfully on a public dataset. The next step is Setting up a Camera so you can run on live data and capture and process your own datasets.

1

https://projects.asl.ethz.ch/datasets/doku.php?id=kmavvisualinertialdatasets

2

https://lifelong-robotic-vision.github.io/dataset/scene

3

https://vision.in.tum.de/data/datasets/visual-inertial-dataset