This script can be used to perform sequenced downloads of various sensors logs for the robotcar-dataset https://robotcar-dataset.robots.ox.ac.uk/.
The code is tested for Python3 on Ubuntu 16.04.
A prebuilt docker image is available with:
docker pull matthewgadd/robotcar-dataset-scraper:latestOtherwise, you can build the docker image from scratch:
wget https://raw.githubusercontent.com/matthewgadd/RobotCarDataset-Scraper/master/Dockerfile
docker build -t matthewgadd/robotcar-dataset-scraper:latest .Mount your downloads dir through the docker image:
docker run --rm -it -w /RobotCarDataset-Scraper -v $HOME/Downloads:/Downloads matthewgadd/robotcar-dataset-scraper:latestThe script
python get_datasets.pyparses the html for the complete dataset listing and scrapes each dataset page and retrieves the available files (tar) for download. With the example output file datasets.csv provided, you should then run:
python scrape_mrgdatashare.py --downloads_dir /Downloads --datasets_file datasets.csv --username USERNAME --password PASSWORDIf you do not need the full dataset, you can use --choice_sensors and --choice_runs_file options to download parts of the dataset on your own needs.
--choice_sensorsoption can receive multiple sensor names intags, stereo_centre, stereo_left, stereo_right, vo, mono_left, mono_right, mono_rear, lms_front, lms_rear, ldmrs, gps, all.--choice_runs_fileoption receive a.txtfile that contains the names of runs you want to download, we provide an example filea sample fileexample_list.txt.
for example you can download "stereo_centre", "vo" and "lms_front" data of "2014-05-19-13-20-57" and "2014-06-26-09-31-18" by the following command:
python scrape_mrgdatashare.py --choice_sensors stereo_centre,vo,lms_front --choice_runs_file example_list.txt --downloads_dir /Downloads --datasets_file datasets.csv --username USERNAME --password PASSWORD