This repository contains bash scripts that use wget to download different parts of the dataset
The repository replicates the folder structure of the dataset
There is one download script for each folder of each level of the directory tree
By running one script, you will download all the content of the folders contained below that node in the directory tree
For example:
download_videos.shwill dowload all the videos in one go.videos/test/download_P01.shwill download only test videos from subject P01frames_rgb_flow/flow/download_train.shwill download all the flow images contained in the train set
To run one script, simply invoke it in any shell where wget is available
The scripts take a single optional parameter, to specify a desired output path
If you do not specify any parameter, the scripts will assume the output path is~/Download
A folder named EPIC_KITCHENS_2018 will be created at the output path as the main dataset folder
The scripts will reconstruct the dataset structure as they are executed, so if you run for example
sh videos/test/download_P16.sh ~/Desktop
the script will create the following folder structure
~/Desktop/EPIC_KITCHENS_2018/videos/test/P16
and will save P16's test videos to the above path
If you will run the script again later using the same output path, already downloaded files will be skipped
You can interrupt any script at any time and resume the download afterwards
Partially downloaded files will be kept and resumed by wget in order to allow you to download big files in multiple goes
Note: These scripts will work for Linux and Mac. For Windows users a bash installation should work