/gamepy

Creates and manages recording games and game inputs. Used to train neural networks that can then spoof human input.

Primary LanguagePythonMIT LicenseMIT

gamepy

Records images and keystrokes during gameplay for use in supervised learning projects. Associates each image of gameplay with a particular keystroke in a csv file.

Dependancies

  • numpy
  • pynput

Usage

The file record.py gives an example of setting up a recording session. The Recorder class captures the screen as an image and associates each image with a monitored key stroke. This produces data that could train a neural network to create a self driving car. The keras_example.py file gives an example of loading such a model that can then be used in conjuction with the KeyController class to produce a self driving car in a simulated gaming environment.