Game_Learner

An ML Project to train the computer to play a game

Milestones

• Select Game to learn from --> HIDDEN GAME ON GOOGLE! (Atari Breakout!). This game will be played using keyboard.
• Extract training data
• Train NN
• Testing - Let it Play

Current status

We have the info about:

• coordinates of the ball and plate.
• key logs and their time intervals during they were held down.

Model

Classification

• Input Vector
  • Time
  • Ball coordinates(b_x,b_y)
  • Plate coordinate(p_x)
• Output Vector
  • Key which was pressed at that time (Key.left, Key.right or None)

Extracting Data from Sample_image.png using OpenCV

Sample Image

After running 1shot_test.py we get the location of ball and plate

Ball with purple boundary Plate with white boudary

Contributers

• Piyush Ranjan - View Contributions
• Piyush Chauhan - View Contributions