基于yolov8的游戏人物自动锁定功能
1️⃣加载模型
2️⃣获取屏幕内容
3️⃣获取人物坐标
4️⃣鼠标移动到指定位置
import math
import random
import time
import mss
from PIL import Image
import pyautogui
import torch
from pynput.mouse import Controller
from ultralytics import YOLO
#import win32api, win32con# 加载本地模型
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = YOLO('yolov8.pt').to(device)
print("Model loaded.")# 定义屏幕宽高
screen_width, screen_height = pyautogui.size()
game_width = screen_width
game_height = screen_heightrect = (0, 0, game_width, game_height)
m = mss.mss()
mouse_controller = Controller() # 初始化鼠标控制器
#Windows
# def move(dx, dy):
# # 鼠标相对当前位置移动
# win32api.mouse_event(win32con.MOUSEEVENTF_MOVE | win32con.MOUSEEVENTF_ABSOLUTE,
# int(dx * 65536 / game_width), int(dy * 65536 / game_height), 0, 0)
#mac
def move(dx, dy):# 鼠标相对当前位置移动pyautogui.moveRel(dx, dy)def window_capture():# 使用mss库进行截图with mss.mss() as sct:sct_img = sct.grab(rect)img = Image.frombytes('RGB', sct_img.size, sct_img.bgra, 'raw', 'BGRX')return img# 主循环
while True:start_time = time.time()try:# 截取屏幕screen_shot = window_capture()# 使用模型进行推理results = model(screen_shot)# 提取检测结果boxes = results[0].boxes.cpu().numpy()# 存储有效检测结果detections = []for box in boxes:class_id = int(box.cls)confidence = float(box.conf)if class_id == 0 and confidence > 0.5:xmin, ymin, xmax, ymax = box.xyxy[0]center_x = int((xmin + xmax) / 2)#center_y = int((ymin + ymax) / 2)center_y = int(ymax-(ymax - ymin) * 0.82)detections.append((center_x, center_y))print(detections)if detections:# 获取鼠标当前位置mouse_x, mouse_y = mouse_controller.position# 计算每个检测结果与鼠标位置的距离，并找到最近的目标distances = [math.sqrt((mouse_x - x) ** 2 + (mouse_y - y) ** 2) for x, y in detections]closest_index = distances.index(min(distances))closest_target = detections[closest_index]# 控制鼠标移动到最近目标处delta_x = closest_target[0] - mouse_xdelta_y = closest_target[1] - mouse_ymove(delta_x, delta_y)except Exception as e:print(f"An error occurred: {e}")elapsed_time = time.time() - start_timeprint(f"Elapsed time: {elapsed_time:.2f} seconds")