/ohwr-datapal

iOS App for creating Handwriting Recognition datasets. (uni/multi-stroke supported)

Primary LanguageSwiftMIT LicenseMIT

Online Handwriting Datapal (iOS)

ipadpreview

Create datasets for online (stroke-based) handwriting recognition / classification tasks.

Features

  • Create your own labels (classes) and stroke samples from your iPad or iPhone (draw by touch)
  • Datasets can be imported / exported to json
  • Preview and edit created samples
  • Single or multistroke samples supported
  • Choose the size of the input canvas
  • Written in SwiftUI

JSON data format

Format of the dataset is the Google QuickDraw json format.

Minimal preview can be found here.

TODO

  • Data persistence (CoreData)
    • this tool was intended for small datasets
    • only a naive data persistence can be achieved in the current version by exporting / importing a json. Use 'Import from JSON' while creating a new dataset.
  • Various export formats / file formats (e.g. csv, IAM style XML)

License

MIT