metadata.json

{
    "description": "This dataset contains original and scribble-processed images.",
    "source": "Source of original images",
    "processing_technique": "Scribble transformation using Processor XYZ",
    "splits": {
      "train": "Number of training images",
      "validation": "Number of validation images",
      "test": "Number of test images"
    }
  }