main.py

# Import environment libraries
from utils import startGameRand, startGameModel
import gym_super_mario_bros
from nes_py.wrappers import JoypadSpace
from gym_super_mario_bros.actions import SIMPLE_MOVEMENT
from utils import SaveOnBestTrainingRewardCallback
from stable_baselines3 import PPO

# Import preprocessing wrappers
from gym.wrappers import GrayScaleObservation
from stable_baselines3.common.vec_env import VecFrameStack, DummyVecEnv, VecMonitor
from matplotlib import pyplot as plt

# Start the environment
env = gym_super_mario_bros.make('SuperMarioBros-v0') # Generates the environment
env = JoypadSpace(env, SIMPLE_MOVEMENT) # Limits the joypads moves with important moves

# Apply the preprocessing
env = GrayScaleObservation(env, keep_dim=True) # Convert to grayscale to reduce dimensionality
env = DummyVecEnv([lambda: env])
# Alternatively, you may use SubprocVecEnv for multiple CPU processors
env = VecFrameStack(env, 4, channels_order='last') # Stack frames
env = VecMonitor(env, "./train/TestMonitor") # Monitor your progress

model = PPO.load('/home/ercihan/Desktop/EE449/HW3/best_model')
startGameModel(env, model)