Scala script to convert Pascal VOC to CSV. For a medium article

voc_to_csv.sc

#!/usr/bin/env -S scala-cli shebang --scala-version 3.2.0

//> using lib "org.scala-lang.modules::scala-xml:2.1.0"

import scala.xml.Elem
import java.io.File
import scala.xml.XML
import java.nio.file.Paths
import scala.util.Using
import java.nio.file.Files

case class Voc(
    filename: String,
    objects: Seq[VocObject] = Nil
)

case class VocObject(
    name: String,
    xmin: Int,
    ymin: Int,
    xmax: Int,
    ymax: Int
):
  val width  = xmax - xmin
  val height = ymax - ymin


object VocParser:

  def parse(xml: Elem): Voc =
    val filename   = (xml \ "filename").text
    val objects    = xml \ "object"
    val vocObjects = objects.map(n => {
      val name = (n \ "name").text
      // VOC is 1-based index. Convert to 0 based.
      // See https://cv.gluon.ai/_modules/gluoncv/data/pascal_voc/detection.html#VOCDetection
      val xmin = (n \ "bndbox" \ "xmin").text.toInt - 1
      val ymin = (n \ "bndbox" \ "ymin").text.toInt - 1
      val xmax = (n \ "bndbox" \ "xmax").text.toInt - 1
      val ymax = (n \ "bndbox" \ "ymax").text.toInt - 1
      VocObject(name, xmin, ymin, xmax, ymax)
    });
    Voc(filename, objects = vocObjects)



if (args.length != 1) {
  println("Usage: voc_to_csv.sc <path to VOC dir>")
  System.exit(1)
}
val dir = Paths.get(args(0))

Using(Files.list(dir)) { stream => 
  stream
    .filter(_.toString.endsWith(".xml"))
    .map(path => XML.loadFile(path.toFile()))
    .map(VocParser.parse)
    .forEach { voc =>
      voc.objects.foreach { obj =>
        println(s"${voc.filename},${obj.name},${obj.xmin},${obj.ymin},${obj.width},${obj.height}")
      }
    }
  }