Argoverse 2 to Foxglove

This project lets you convert the Argoverse 2 dataset to MCAP to be visualized in Foxglove.

To get started, clone this repository and pip install the dependencies in requirements.txt.
Then download the Argoverse dataset by following this guide (specifically, download the logs found in "s3://argoverse/datasets/av2/sensor/train").

To convert a log to MCAP, execute the following command

python3 ./foxglove-av2/convert.py --data_root <path to dataset dir> --log_id <log id>

For example

python3 ./foxglove-av2/convert.py --data_root /home/alp/data/datasets/ --log_id 00a6ffc1-6ce9-3bc3-a060-6006e9893a1a

This will produce an MCAP file for each type of data (annotations, camera, lidar, pose).
To combine the files into a single MCAP file, download the MCAP command line tool and use the merge command.