# window_size = (0,0,1920,1017)
# station_size = (230, 230, 1670, 930)
# WIDTH = 400
# HEIGHT = 200
# hp_station = cv2.cvtColor(cv2.resize(grab_screen(station_size),(WIDTH,HEIGHT)),cv2.COLOR_BGR2GRAY)
# # boss_blood = boss_hp(hp_station, 570)
# # last_hp = boss_blood
# # next_self_blood = player_hp(hp_station)
# min_hp = 9
# check_point = (612, 187)
# # start_time = time.time()
h = Hp_getter()
last_hy = 0
while True:
# take_action(6)
px, py = h.get_play_location()
hx, hy = h.get_hornet_location()
if last_hy > 32 and last_hy < 32.5 and hy > 32 and last_hy < 32.5:
print("skill")
last_hy = hy
time.sleep(0.25)
# print(direction_reward(0, px, hx), " ",distance_reward(0, px, hx), " ", px - hx)
# print(h.get_play_location(), " ",h.get_hornet_location())
# hp_station = cv2.cvtColor(cv2.resize(grab_screen(station_size),(WIDTH,HEIGHT)),cv2.COLOR_RGBA2RGB)
# fn = "./test_img/" + str(i) + ".png"
# cv2.imwrite(fn, hp_station)
# time.sleep(0.02)
total_remind_hp = 0
act_rmp_correct = ReplayMemory(
MEMORY_SIZE, file_name='./act_correct_memory') # experience pool
act_rmp_wrong = ReplayMemory(
MEMORY_SIZE, file_name='./act_wrong_memory') # experience pool
move_rmp_correct = ReplayMemory(
MEMORY_SIZE, file_name='./move_correct_memory') # experience pool
move_rmp_wrong = ReplayMemory(
MEMORY_SIZE, file_name='./move_wrong_memory') # experience pool
# new model, if exit save file, load it
model = Model(INPUT_SHAPE, ACTION_DIM)
# Hp counter
hp = Hp_getter()
model.load_model()
algorithm = DQN(model, gamma=GAMMA, learnging_rate=LEARNING_RATE)
agent = Agent(ACTION_DIM, algorithm, e_greed=0, e_greed_decrement=1e-6)
# get user input, no need anymore
# user = User()
# paused at the begining
paused = True
paused = Tool.Helper.pause_game(paused)
max_episode = 30000
# 开始训练
episode = 0