Dataset

This layer (dataset) places every image that it sees to dataset with a specified name.
Put name of the future dataset in the field name.
{
  "action": "dataset",
  "src": ["$input" ],
  "dst": "$output",
  "settings": {
    "name": "new_dataset_name"
  }
}
Special case: preserve original dataset names
{
  "action": "dataset",
  "src": ["$input" ],
  "dst": "$output",
  "settings": {
    "rule": "save_original"
  }
}
If "rule": "save_original" passed instead of name then each image will be stored in dataset named like original dataset in source project for the image.
Example
In this example we will put images that have tag "good" to separate dataset good and keep other images in default dataset (named like "SourceProject__SourceDataset").
Input project
Output project
[
  {
    "dst": "$data",
    "src": [
      "example/*"
    ],
    "action": "data",
    "settings": {
      "classes_mapping": "default"
    }
  },
  {
    "action": "if",
    "src": ["$data" ],
    "dst": [
      "$good",
      "$bad"
    ],
    "settings": {
      "condition": {
        "tags": ["good"]
      }
    }
  },
  {
    "action": "dataset",
    "src": ["$good" ],
    "dst": "$good_ds",
    "settings": {
      "name": "good"
    }
  },
  {
    "dst": "example_dataset",
    "src": [
      "$good_ds",
      "$bad"
    ],
    "action": "supervisely",
    "settings": {}
  }
]
Crop
Drop by Class
Last updated 5 months ago