This repository can be used as a tool to extract information from frames of any game which can otherwise not be used for research purpose.
Steps:
- Acquire a video of the game (make sure resolution of this video is the same as that of actual gameplay)
- Find the coordinates of region of interests (ROIs) using utils
- Use the generated
key_info.log
actual gameplay video as input formain.py
to extract values written in ROI
These extracted values for example can be used as reward values for training any model.
python3 main.py --video_path game.mp4 --key_info ./key_info.log --model Model/model.pyt
- Data collector to fine tune neural network for any given game
- Add information about hacks to control game in order to automate keyboard inputs