import.md

Import CLI

If you use Mass Labeling within the docker container, don't forget to put data inside the container. It can be done using the data folder, which is mounted to the container by default. The import command should be run inside the container.

Import dataset

Run command

bin/cli import:dataset <arguments> | npx bunyan

Arguments

• --in - path to the zipped dataset file. Default: data/import/dataset.zip. And the default could be changed under the System tab in the Setting menu.

An input file is a zip archive file which contains the following:

• screenshots
  • 0123456789abcdef01234567.jpg
  • ...
• description.json

screenshots folder contains JPEG images which are going to be shown to the assessors. description.json file consists of the item array. Each item satisfies the following format:

{
    "url"           : str,      // "http://site.com"
    "dataset"       : str,      // "my_dataset"
    "screenshot"    : str,      // "screenshots/0123456789abcdef01234567.jpg"
}

Import rates

You can import rates from outside (sources other than Mass Labeling) to use them in sliders (see the slider description in the user guide).

Run command

bin/cli import:modelRates <arguments> | npx bunyan

Arguments

• --in - path to the file with rates. Default: data/import/modelRates.json. And the default could be changed under the System tab in the Setting menu.

An input file is a json file which contains the item array. Each item satisfies the following format:

{
    "id"            : str       // "http://indastro.ru/en",
    "score"         : int       // 1
}