/deep-rps

rock-paper-scissors game with deep learning

Primary LanguageScalaMIT LicenseMIT

deep-rps

rock-paper-scissors game with deep learning

Build Status CircleCI

app

alt text

features

  1. train neural network (fed with images)
  2. auto save/load network model to/from ~/.deep-rps/model.zip
  3. evaluate images (works with loaded model)
  4. play rock-paper-scissors game (works with loaded model)

game features

  1. player vs bot
  2. you are the player
  3. random choice for bot
  4. play without webcam (select hand on static image)
  5. play with webcam
  6. live network evaluation

--help

deep-rps 0.0.2
Usage: deep-rps [options]

  -x, --x-gui        run game with GUI
  -e, --eval <file>  image to eval (can be used few times to provide multiple files)
  -t, --train <dir>  train network with provided dir with train and test subdirs
  --help             prints this usage text
  --version          prints app version

maybe in future

  1. object detection
  2. tool to prepare train/test images