1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
| import sys
import time
import torch
import win32api
import win32con
from PyQt5.QtWidgets import QApplication
GAME_WIDTH = 1920
GAME_HEIGHT = 1080
IMG_PATH = 'screenshot.bmp'
HUBCONF_FOLDER_PATH = r'.'
MODEL_PATH = 'best.pt'
CONFIDENCE_THRESHOLD = 0.4
MOVE_DELAY1 = 0.049
MOVE_DELAY2 = 0.033
ONE_SHOT_DELAY = 0.001
def screenshot(screen, img_path):
img = screen.grabWindow(0).toImage()
img.save(img_path)
def inference_by_image(model, img_path):
results = model(img_path)
results_pandas = results.pandas()
x_mins = results_pandas.xyxy[0]['xmin']
x_maxs = results_pandas.xyxy[0]['xmax']
y_mins = results_pandas.xyxy[0]['ymin']
y_maxs = results_pandas.xyxy[0]['ymax']
confidences = results_pandas.xyxy[0]['confidence']
return x_mins, x_maxs, y_mins, y_maxs, confidences
def get_points_by_threshold(zip_object, threshold):
points = []
for x_min, x_max, y_min, y_max, conf in zip_object:
if conf > threshold:
points.append([int(x_min), int(x_max), int(y_min), int(y_max)])
return points
def get_center_pos(points):
center_pos = []
for point in points:
x_center = int(point[1] - (point[1] - point[0]) // 2)
y_center = int(point[3] - (point[3] - point[2]) // 2)
center_pos.append([x_center, y_center])
return center_pos
def get_move_offsets(center_pos, game_width, game_height):
move_offsets = []
for center in center_pos:
x_offset = int(center[0] - game_width // 2)
y_offset = int(center[1] - game_height // 2)
move_offsets.append([x_offset, y_offset])
return move_offsets
def fine_tuning(x_offset, y_offset):
if x_offset > 0:
x_offset += 43
if y_offset > 0:
y_offset += 44
if x_offset < 0:
x_offset -= 37
if y_offset < 0:
y_offset -= 44
if y_offset < -200:
y_offset -= 45
elif y_offset < -150:
y_offset -= 30
return x_offset, y_offset
def move_and_click_by_offsets(x_offset, y_offset, move_delay1, move_delay2):
win32api.mouse_event(win32con.MOUSEEVENTF_MOVE, x_offset, y_offset, 0, 0)
time.sleep(move_delay1)
win32api.mouse_event(win32con.MOUSEEVENTF_LEFTDOWN, 0, 0, 0, 0)
win32api.mouse_event(win32con.MOUSEEVENTF_LEFTUP, 0, 0, 0, 0)
time.sleep(move_delay2)
win32api.mouse_event(win32con.MOUSEEVENTF_MOVE, -x_offset, -y_offset, 0, 0)
def main():
device = torch.device('cuda')
model = torch.hub.load(HUBCONF_FOLDER_PATH, 'custom', MODEL_PATH,
source='local', force_reload=False)
model = model.to(device)
app = QApplication(sys.argv)
screen = QApplication.primaryScreen()
while True:
screenshot(screen, IMG_PATH)
x_mins, x_maxs, y_mins, y_maxs, confidences = inference_by_image(model, IMG_PATH)
points = get_points_by_threshold(zip(x_mins, x_maxs, y_mins, y_maxs, confidences),
CONFIDENCE_THRESHOLD)
if len(points) == 3:
center_pos = get_center_pos(points)
move_offsets = get_move_offsets(center_pos, GAME_WIDTH, GAME_HEIGHT)
for x_offset, y_offset in move_offsets:
x_offset, y_offset = fine_tuning(x_offset, y_offset)
move_and_click_by_offsets(x_offset, y_offset, MOVE_DELAY1, MOVE_DELAY2)
time.sleep(ONE_SHOT_DELAY)
if __name__ == '__main__':
main()
|
