PongGameRL

A reinforcement learning environment for automating and training agents to play Pong using screen capture and keyboard control.

Features

• Uses OpenAI Gymnasium interface for RL compatibility.
• Captures game frames using mss and processes them with cv2.
• Controls the game via pyautogui for mouse and keyboard automation.
• Detects game end state using OCR (pytesseract) on winner/loser text.
• Includes helper functions for mouse position and image display.

Requirements

• Python 3.10.13 (use a virtual environment)
• mss
• opencv-python
• pyautogui
• numpy
• pytesseract
• matplotlib
• gymnasium
• stable-baselines3

Install dependencies:

pip install mss opencv-python pyautogui numpy pytesseract matplotlib gymnasium stable-baselines3