yolov5_d435i_detection-main

YOLOV3_Fire_Detection-yolo编程资源

YOLOV3_Fire_Detection-yolo编程资源是一个专注于使用YOLOv3算法进行火灾检测的编程项目资源包。YOLOv3是一种流行的目标检测算法，具有速度快，准确度高，实时性好的特点。在此项目中，它被应用于火灾检测领域，能够...

部署_YOLOv5_detection_with_Streamlit_YOLOv5-Streamlit_Deploy

部署_YOLOv5_detection_with_Streamlit_YOLOv5-Streamlit_Deploy_YOLOv5_detection_with_Streamlit_yolov5-streamlit

基于Intel_Realsense_D435i深度相机的YOLOv11实时目标检测与三维定位系统_使用PyTorch框架实现高效深度学习模型_支持80类COCO数据集目标识别_结合.zip

yolov11_d435i_detection-main文件夹可能包含了源代码、训练好的模型文件以及必要的配置文件。基于Intel Realsense D435i深度相机和YOLOv11算法的实时目标检测与三维定位系统，在PyTorch框架的加持下，支持COCO...

YOLOv5_DOTA_OBB-master：Windows环境中的性能优化技巧，提升部署效率

本文介绍了YOLOv5_DOTA_OBB的集成、性能分析和优化，并探讨了其在Windows环境中的部署效率提升。首先，概述了YOLOv5_DOTA_OBB项目的结构和特点，并说明了Windows环境搭建的具体要求及步骤。接着，文章详细分析了性能...

Traceback (most recent call last): File "D:\pythonproject36\yolov5-master_hand_detection\train.py", line 635, in <module> main(opt) File "D:\pythonproject36\yolov5-master_hand_detection\train.py", line 520, in main device = select_device(opt.device, batch_size=opt.batch_size) File "D:\pythonproject36\yolov5-master_hand_detection\utils\torch_utils.py", line 75, in select_device p = torch.cuda.get_device_properties(i) File "D:\Program Files (x86)\Anaconda2023\envs\deep\lib\site-packages\torch\cuda\init.py", line 398, in get_device_properties raise AssertionError("Invalid device id") AssertionError: Invalid device id

这个错误是由于在选择设备时，传入了无效的设备ID引起的。请确保你传入的设备ID是有效的。你可以尝试以下解决方法： 1. 检查你的设备ID是否正确。可以通过 torch.cuda.device_count() 来获取可用的GPU数量，并...

",line 635, in <module> main(opt) File "D:\pythonproject36\yolov5-master_hand_detection\train.py", line 520, in main device = select_device(opt.device, batch_size=opt.batch_size) File "D:\pythonproject36\yolov5-master_hand_detection\utils\torch_utils.py", line 75, in select_device p = torch.cuda.get_device_properties(i) File "D:\Program Files (x86)\Anaconda2023\envs\deep\lib\site-packages\torch\cuda\init.py", line 398, in get_device_properties raise AssertionError("Invalid device id") AssertionError: Invalid device id

这个错误是由于使用了无效的设备id导致的。在你的代码中，调用了select_device函数，该函数使用了torch.cuda.get_device_properties来获取设备属性。然而，传递给select_device函数的设备id无效，导致引发了...

Yolov5与Intel Realsense D435i物体三维距离测量工具包

例如，用户需要配置 yolov5_D435i_2.0/config/yolov5s.yaml 来设置 Yolov5 的参数，运行 yolov5_D435i_2.0/config/main2.py 来执行三维距离测量的程序。为了确保系统的稳定运行，用户还应注意硬件连接和分辨率参数的...

Handling Class Imbalance in YOLOv8 Object Detection Tasks

# 1. Overview of the Class Imbalance Problem The class imbalance problem is prevalent in machine learning, referring to an uneven distribution of sample quantities across different classes within a ...

Traceback (most recent call last): File "D:/YOLOV8/ultralytics-main/datasets/demo_2.py", line 10, in <module> results = model.train(data='data.yaml', epochs=100,workers=0,batch=16,imgsz=640,device=0, lr0= 0.001,weight_decay=0.05,mixup=0.2, hsv_h=0.015) File "D:\YOLOV8\ultralytics-main\ultralytics\engine\model.py", line 803, in train self.trainer.train() File "D:\YOLOV8\ultralytics-main\ultralytics\engine\trainer.py", line 207, in train self._do_train(world_size) File "D:\YOLOV8\ultralytics-main\ultralytics\engine\trainer.py", line 385, in _do_train self.loss, self.loss_items = self.model(batch) File "D:\anaconda\envs\yolo8\lib\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl return forward_call(*input, **kwargs) File "D:\YOLOV8\ultralytics-main\ultralytics\nn\tasks.py", line 111, in forward return self.loss(x, *args, **kwargs) File "D:\YOLOV8\ultralytics-main\ultralytics\nn\tasks.py", line 293, in loss return self.criterion(preds, batch) File "D:\YOLOV8\ultralytics-main\ultralytics\utils\loss.py", line 268, in call loss[0], loss[2] = self.bbox_loss( File "D:\anaconda\envs\yolo8\lib\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl return forward_call(*input, **kwargs) File "D:\YOLOV8\ultralytics-main\ultralytics\utils\loss.py", line 128, in forward loss_cls = self.focal_loss(pred_cls, target_cls).sum() / target_scores_sum File "D:\anaconda\envs\yolo8\lib\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl return forward_call(*input, **kwargs) File "D:\YOLOV8\ultralytics-main\ultralytics\utils\loss.py", line 50, in forward loss = F.binary_cross_entropy_with_logits(pred, label, reduction="none") File "D:\anaconda\envs\yolo8\lib\site-packages\torch\nn\functional.py", line 2980, in binary_cross_entropy_with_logits raise ValueError("Target size ({}) must be the same as input size ({})".format(target.size(), input.size())) ValueError: Target size (torch.Size([793, 5])) must be the same as input size (torch.Size([793, 64]))帮我修改class FocalLoss(nn.Module): """Wraps focal loss around existing loss_fcn(), i.e. criteria = FocalLoss(nn.BCEWithLogitsLoss(), gamma=1.5).""" def init(self, alpha=3.0, gamma=2): # 添加参数 super().init() self.alpha = alpha self.gamma = gamma @staticmethod def forward(pred, label, gamma=2, alpha=3.0): """Calculates and updates confusion matrix for object detection/classification tasks.""" loss = F.binary_cross_entropy_with_logits(pred, label, reduction="none") # p_t = torch.exp(-loss) # loss = self.alpha (1.000001 - p_t) ** self.gamma # non-zero power for gradient stability # TF implementation https://github.com/tensorflow/addons/blob/v0.7.1/tensorflow_addons/losses/focal_loss.py pred_prob = pred.sigmoid() # prob from logits p_t = label * pred_prob + (1 - label) * (1 - pred_prob) modulating_factor = (1.0 - p_t) ** gamma loss = modulating_factor if alpha > 0: alpha_factor = label alpha + (1 - label) * (1 - alpha) loss *= alpha_factor return loss.mean(1).sum()

嗯，我现在需要帮用户解决这个YOLOv8训练时的错误。错误信息是ValueError，提示目标尺寸和输入尺寸不一致，具体是Target size (torch.Size([793, 5]))和输入尺寸(torch.Size([793, 64]))不匹配。用户提供了他们的...

File "D:\lab\yolov12-main\yolov12-main\ultralytics\models\yolo\detect\val.py", line 181, in get_stats stats = {k: torch.cat(v, 0).cpu().numpy() for k, v in self.stats.items()} # to numpy ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "D:\lab\yolov12-main\yolov12-main\ultralytics\models\yolo\detect\val.py", line 181, in <dictcomp> stats = {k: torch.cat(v, 0).cpu().numpy() for k, v in self.stats.items()} # to numpy ^^^^^^^^^^^^^^^ RuntimeError: torch.cat(): expected a non-empty list of Tensors

好的，我现在需要帮助用户解决在使用YOLOv12进行目标检测时出现的RuntimeError: torch.cat()接收到空列表的问题。用户提到在运行val.py时出现这个错误，并且引用了之前的回答和站内引用，我需要结合这些信息来给出...

from ultralytics import YOLO import cv2 # 1. 加载预训练模型权重 model = YOLO("D:/YOLOV8/ultralytics-main/datasets/runs/detect/train8/weights/best.pt") # 替换为你的权重文件路径[^3] # 2. 读取待检测图像 image_path = "D:/YOLOV8/ultralytics-main/datasets/lytdatasets/divide-MT/valid/images/RRT-13R_Img1_A80_S3_[10][70].png" image = cv2.imread(image_path) # 使用OpenCV读取图像[^1] # 3. 执行推理预测 results = model.predict(image, save=True, conf=0.5) # 设置置信度阈值并自动保存结果[^2] # 4. 保存检测结果文本信息 for i, result in enumerate(results): # 获取检测结果DataFrame df = result.pandas().xyxy[0] # 保存检测结果到CSV文件 df.to_csv(f"detection_result_{i}.csv", index=False) # 打印检测结果摘要 print(f"图像{i + 1}检测到{len(df)}个目标：") print(df[["name", "confidence", "xmin", "ymin", "xmax", "ymax"]])

好的，用户的问题是关于如何使用YOLOv8加载自定义权重进行图像检测并将结果保存为CSV。我需要先理清YOLOv8的基本使用流程，然后结合用户提到的自定义权重和CSV保存来构建步骤。首先，用户提到了加载自定义权重文件...

Traceback (most recent call last): File "D:\桌面\ultralytics-yolov8-main\test1\blueberry.py", line 4, in <module> a1 = YOLO('yolov8n.pt') # 官方提供的基础测试和训练模型 ^^^^^^^^^^^^^^^^^^ File "D:\桌面\ultralytics-yolov8-main\ultralytics\yolo\engine\model.py", line 106, in init self._load(model, task) File "D:\桌面\ultralytics-yolov8-main\ultralytics\yolo\engine\model.py", line 145, in _load self.model, self.ckpt = attempt_load_one_weight(weights) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "D:\桌面\ultralytics-yolov8-main\ultralytics\nn\tasks.py", line 396, in attempt_load_one_weight ckpt, weight = torch_safe_load(weight) # load ckpt ^^^^^^^^^^^^^^^^^^^^^^^ File "D:\桌面\ultralytics-yolov8-main\ultralytics\nn\tasks.py", line 336, in torch_safe_load return torch.load(file, map_location='cpu'), file # load ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\李佳伟\AppData\Roaming\Python\Python312\site-packages\torch\serialization.py", line 1524, in load raise pickle.UnpicklingError(_get_wo_message(str(e))) from None _pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. (1) In PyTorch 2.6, we changed the default value of the weights_only argument in torch.load from False to True. Re-running torch.load with weights_only set to False will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source. (2) Alternatively, to load with weights_only=True please check the recommended steps in the following error message. WeightsUnpickler error: Unsupported global: GLOBAL ultralytics.nn.tasks.DetectionModel was not an allowed global by default. Please use torch.serialization.add_safe_globals([ultralytics.nn.tasks.DetectionModel]) or the torch.serialization.safe_globals([ultralytics.nn.tasks.DetectionModel]) context manager to allowlist this global if you trust this class/function. Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.解释代码

具体而言就是先借助常规 I/O 流程获取原始字节串然后再交由专门负责此项工作的类方法去进一步加工整理成最终期望的形式。例如下面这段伪代码展示了大致思路： python with open(model_path, 'rb') as fobj: ...

import os import cv2 import sys import argparse import time # add path realpath = os.path.abspath(file) _sep = os.path.sep realpath = realpath.split(_sep) try: zoo_root_index = next(i for i, part in enumerate(realpath) if 'rknn_model_zoo' in part) rknn_model_zoo_path = os.path.join(realpath[0]+_sep, realpath[1:zoo_root_index+1]) sys.path.append(rknn_model_zoo_path) except StopIteration: raise ValueError("Could not find 'rknn_model_zoo' directory in the path: {}".format(os.path.abspath(file))) from py_utils.coco_utils import COCO_test_helper import numpy as np OBJ_THRESH = 0.25 NMS_THRESH = 0.45 # The follew two param is for map test # OBJ_THRESH = 0.001 # NMS_THRESH = 0.65 IMG_SIZE = (640, 640) # (width, height), such as (1280, 736) CLASSES = ("car","white") coco_id_list = [1,2] def filter_boxes(boxes, box_confidences, box_class_probs): """Filter boxes with object threshold. """ box_confidences = box_confidences.reshape(-1) candidate, class_num = box_class_probs.shape class_max_score = np.max(box_class_probs, axis=-1) classes = np.argmax(box_class_probs, axis=-1) _class_pos = np.where(class_max_score box_confidences >= OBJ_THRESH) scores = (class_max_score* box_confidences)[_class_pos] boxes = boxes[_class_pos] classes = classes[_class_pos] return boxes, classes, scores def nms_boxes(boxes, scores): """Suppress non-maximal boxes. # Returns keep: ndarray, index of effective boxes. """ x = boxes[:, 0] y = boxes[:, 1] w = boxes[:, 2] - boxes[:, 0] h = boxes[:, 3] - boxes[:, 1] areas = w * h order = scores.argsort()[::-1] keep = [] while order.size > 0: i = order[0] keep.append(i) xx1 = np.maximum(x[i], x[order[1:]]) yy1 = np.maximum(y[i], y[order[1:]]) xx2 = np.minimum(x[i] + w[i], x[order[1:]] + w[order[1:]]) yy2 = np.minimum(y[i] + h[i], y[order[1:]] + h[order[1:]]) w1 = np.maximum(0.0, xx2 - xx1 + 0.00001) h1 = np.maximum(0.0, yy2 - yy1 + 0.00001) inter = w1 * h1 ovr = inter / (areas[i] + areas[order[1:]] - inter) inds = np.where(ovr <= NMS_THRESH)[0] order = order[inds + 1] keep = np.array(keep) return keep def dfl(position): # Distribution Focal Loss (DFL) import torch x = torch.tensor(position) n,c,h,w = x.shape p_num = 4 mc = c//p_num y = x.reshape(n,p_num,mc,h,w) y = y.softmax(2) acc_metrix = torch.tensor(range(mc)).float().reshape(1,1,mc,1,1) y = (yacc_metrix).sum(2) return y.numpy() def box_process(position): grid_h, grid_w = position.shape[2:4] col, row = np.meshgrid(np.arange(0, grid_w), np.arange(0, grid_h)) col = col.reshape(1, 1, grid_h, grid_w) row = row.reshape(1, 1, grid_h, grid_w) grid = np.concatenate((col, row), axis=1) stride = np.array([IMG_SIZE[1]//grid_h, IMG_SIZE[0]//grid_w]).reshape(1,2,1,1) position = dfl(position) box_xy = grid +0.5 -position[:,0:2,:,:] box_xy2 = grid +0.5 +position[:,2:4,:,:] xyxy = np.concatenate((box_xystride, box_xy2stride), axis=1) return xyxy def post_process(input_data): boxes, scores, classes_conf = [], [], [] defualt_branch=3 pair_per_branch = len(input_data)//defualt_branch # Python 忽略 score_sum 输出 for i in range(defualt_branch): boxes.append(box_process(input_data[pair_per_branchi])) classes_conf.append(input_data[pair_per_branchi+1]) scores.append(np.ones_like(input_data[pair_per_branchi+1][:,:1,:,:], dtype=np.float32)) def sp_flatten(_in): ch = _in.shape[1] _in = _in.transpose(0,2,3,1) return _in.reshape(-1, ch) boxes = [sp_flatten(_v) for _v in boxes] classes_conf = [sp_flatten(_v) for _v in classes_conf] scores = [sp_flatten(_v) for _v in scores] boxes = np.concatenate(boxes) classes_conf = np.concatenate(classes_conf) scores = np.concatenate(scores) # filter according to threshold boxes, classes, scores = filter_boxes(boxes, scores, classes_conf) # nms nboxes, nclasses, nscores = [], [], [] for c in set(classes): inds = np.where(classes == c) b = boxes[inds] c = classes[inds] s = scores[inds] keep = nms_boxes(b, s) if len(keep) != 0: nboxes.append(b[keep]) nclasses.append(c[keep]) nscores.append(s[keep]) if not nclasses and not nscores: return None, None, None boxes = np.concatenate(nboxes) classes = np.concatenate(nclasses) scores = np.concatenate(nscores) return boxes, classes, scores def draw(image, boxes, scores, classes): for box, score, cl in zip(boxes, scores, classes): top, left, right, bottom = [int(_b) for _b in box] print("%s @ (%d %d %d %d) %.3f" % (CLASSES[cl], top, left, right, bottom, score)) cv2.rectangle(image, (top, left), (right, bottom), (255, 0, 0), 2) cv2.putText(image, '{0} {1:.2f}'.format(CLASSES[cl], score), (top, left - 6), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2) def setup_model(args): model_path = args.model_path if model_path.endswith('.pt') or model_path.endswith('.torchscript'): platform = 'pytorch' from py_utils.pytorch_executor import Torch_model_container model = Torch_model_container(args.model_path) elif model_path.endswith('.rknn'): platform = 'rknn' from py_utils.rknn_executor import RKNN_model_container model = RKNN_model_container(args.model_path, args.target, args.device_id) elif model_path.endswith('onnx'): platform = 'onnx' from py_utils.onnx_executor import ONNX_model_container model = ONNX_model_container(args.model_path) else: assert False, "{} is not rknn/pytorch/onnx model".format(model_path) print('Model-{} is {} model, starting val'.format(model_path, platform)) return model, platform def img_check(path): img_type = ['.jpg', '.jpeg', '.png', '.bmp'] for _type in img_type: if path.endswith(_type) or path.endswith(_type.upper()): return True return False if name == 'main': # Create a dummy args object class Args: pass args = Args() args.model_path = '/home/cat/NPU/rknn_model_zoo-main/examples/yolov8/model/whitenu8.rknn' args.target = 'rk3576' args.device_id = None # init model model, platform = setup_model(args) co_helper = COCO_test_helper(enable_letter_box=True) # init camera cap = cv2.VideoCapture(0) if not cap.isOpened(): print("Error: Could not open camera.") exit() print("Press 'q' to quit.") # run test while True: start_time = time.time() ret, img_src = cap.read() if not ret: print("Error: Failed to capture frame.") break # Due to rga init with (0,0,0), we using pad_color (0,0,0) instead of (114, 114, 114) pad_color = (0,0,0) img = co_helper.letter_box(im= img_src.copy(), new_shape=(IMG_SIZE[1], IMG_SIZE[0]), pad_color=(0,0,0)) img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB) # preprocee if not rknn model if platform in ['pytorch', 'onnx']: input_data = img.transpose((2,0,1)) input_data = input_data.reshape(1,*input_data.shape).astype(np.float32) input_data = input_data/255. else: input_data = img outputs = model.run([input_data]) boxes, classes, scores = post_process(outputs) img_p = img_src.copy() if boxes is not None: draw(img_p, co_helper.get_real_box(boxes), scores, classes) end_time = time.time() fps = 1 / (end_time - start_time) cv2.putText(img_p, f"FPS: {int(fps)}", (10, 30), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2) cv2.imshow("yolov8 detection", img_p) if cv2.waitKey(1) & 0xFF == ord('q'): break # release cap.release() cv2.destroyAllWindows() model.release()

我们正在讨论如何调试和优化使用RKNN模型和OpenCV的YOLOv8目标检测代码。由于用户提到在运行YOLOv8模型时遇到问题，并且希望使用RKNN模型（Rockchip Neural Network，一种用于Rockchip芯片的深度学习推理框架）和...

(yolov11) PS D:\deepsort_YOLO11_pytorch-main> python -c "import torch; print(torch.cuda.is_available())" OMP: Error #15: Initializing libiomp5md.dll, but found libiomp5md.dll already initialized. OMP: Hint This means that multiple copies of the OpenMP runtime have been linked into the program. That is dangerous, since it can degrade performance or cause incorrect results. The best thing to do i s to ensure that only a single OpenMP runtime is linked into the process, e.g. by avoiding static linking of the OpenMP runtime in any library. As an unsafe, unsupported, undocumented workaround you ca n set the environment variable KMP_DUPLICATE_LIB_OK=TRUE to allow the program to continue to execute, but that may cause crashes or silently produce incorrect results. For more information, please see http://www.intel.com/software/products/support/.

引用[5]虽然主要讲MMDetection的问题，但同样涉及到模块初始化的问题，可能与dll冲突有关，但此处可以忽略，因为用户的问题更直接与OpenMP相关。接下来，我需要整理步骤，确保用户能够逐步操作。首先推荐临时解决...

import rclpy from rclpy.node import Node from sensor_msgs.msg import CompressedImage from geometry_msgs.msg import Twist from cv_bridge import CvBridge import cv2 import numpy as np import time import math from test1_yolo_bin import YoloNV12Inferencer class LineAndObstacleNode(Node): def init(self): super().init('line_and_obstacle_node') self.bridge = CvBridge() self.subscription = self.create_subscription( CompressedImage, '/image', self.image_callback, 10) self.publisher_ = self.create_publisher(Twist, '/cmd_vel', 10) # YOLO 模型加载 try: self.inferencer = YoloNV12Inferencer('/root/dev_ws/3w/bin/yolov5s_672x672_nv12.bin') self.get_logger().info("✅ YOLO模型加载成功") except Exception as e: self.get_logger().error(f"YOLO加载失败: {e}") self.inferencer = None # 避障参数 self.focal_length = 550.0 self.image_cx = 320.0 self.target_height = 0.3 self.repulsion_range = 0.5 self.state = 'normal' # normal / avoiding_turn / avoiding_straight / realignment self.avoid_start_time = 0 self.obstacle_list = [] # 新增：航向角记录 self.target_yaw = 0.0 # 避障前原始航向 self.current_yaw = 0.0 # 当前航向（简化的航向估计） self.last_yaw_update = time.time() # PID控制参数（回正阶段使用） self.Kp = 1.5 # 比例系数 self.Ki = 0.1 # 积分系数 self.Kd = 0.3 # 微分系数 self.prev_error = 0.0 self.integral = 0.0 # 车道线检测参数 self.min_speed = 0.25 self.lost_lane_counter = 0 self.lost_lane_threshold = 5 self.current_frame = None self.last_valid_alpha = 0.0 # 状态超时参数 self.REALIGN_TIMEOUT = 3.0 # 回正阶段最大超时时间 self.AVOID_TURN_DURATION = 0.6 # 避障转向持续时间 self.YAW_THRESHOLD = 0.1 # 航向角误差阈值（弧度） self.timer = self.create_timer(0.1, self.control_loop) def image_callback(self, msg): try: img = self.bridge.compressed_imgmsg_to_cv2(msg, desired_encoding='bgr8') # 保存原始图像用于YOLO检测 self.raw_image = img.copy() # --- 巡线区域处理 --- h1, width = img.shape[:2] crop = img[int(h1 * 0.4):, :] self.current_frame = cv2.resize(crop, (160, 120)) # --- 简化的航向估计（基于最后有效转向角）--- # 在实际应用中应使用IMU数据 current_time = time.time() time_diff = current_time - self.last_yaw_update if time_diff > 0: # 角速度 = 转向角 / 时间 angular_velocity = self.last_valid_alpha / time_diff self.current_yaw += angular_velocity * time_diff # 角度归一化 [-π, π] self.current_yaw = self.normalize_angle(self.current_yaw) self.last_yaw_update = current_time # --- YOLO处理 --- if self.inferencer: detections = self.inferencer.infer(self.raw_image) self.obstacle_list = [] for class_id, score, (x1, y1, x2, y2) in detections: h = y2 - y1 if h <= 0: continue distance = (self.target_height * self.focal_length) / h cx = (x1 + x2) / 2 offset_y = (cx - self.image_cx) * distance / self.focal_length if class_id == 0: # 假设0是障碍物类别 self.obstacle_list.append((distance, offset_y)) except Exception as e: self.get_logger().error(f"图像处理错误: {e}") def normalize_angle(self, angle): """将角度归一化到[-π, π]范围""" while angle > math.pi: angle -= 2 * math.pi while angle < -math.pi: angle += 2 * math.pi return angle def pid_control(self, target, current): """PID控制器计算角速度""" error = self.normalize_angle(target - current) # PID计算 self.integral += error * 0.1 # 0.1是控制周期 derivative = (error - self.prev_error) / 0.1 # 抗积分饱和 self.integral = max(min(self.integral, 1.0), -1.0) # 计算输出 output = self.Kp * error + self.Ki * self.integral + self.Kd * derivative self.prev_error = error # 输出限制 return max(min(output, 1.0), -1.0) def detect_lane_line(self, frame): """检测车道线并返回速度、转向角和检测状态""" if frame is None: return 0.0, 0.0, False try: hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV) # 黑色线条 HSV 阈值 lower_black = np.array([0, 0, 0]) upper_black = np.array([180, 255, 50]) black_mask = cv2.inRange(hsv, lower_black, upper_black) # 形态学操作增强检测 kernel = np.ones((5, 5), np.uint8) black_mask = cv2.morphologyEx(black_mask, cv2.MORPH_OPEN, kernel) black_mask = cv2.dilate(black_mask, kernel, iterations=2) # 检测区域高度 detection_band = 40 x_coords = np.where(black_mask[-detection_band:] == 255)[1] if len(x_coords) > 50: # 确保有足够的像素点 self.lost_lane_counter = 0 mean_x = np.mean(x_coords) h, w = frame.shape[:2] alpha = float(-(mean_x - w / 2) / (w / 2)) self.last_valid_alpha = alpha # 保存有效转向角 return 0.25, alpha, True else: self.lost_lane_counter += 1 if self.lost_lane_counter >= self.lost_lane_threshold: return 0.0, 0.0, False else: # 使用最后有效的转向角 return 0.10, self.last_valid_alpha, False except Exception as e: self.get_logger().error(f"车道线检测错误: {e}") return 0.0, 0.0, False def control_loop(self): now = time.time() elapsed_time = now - self.avoid_start_time # 障碍物检测（仅考虑前方0.5米内的障碍物） valid_obs = [o for o in self.obstacle_list if 0 < o[0] < self.repulsion_range and abs(o[1]) < 0.5] # 状态机实现 if self.state == 'avoiding_turn': # 右转避障阶段 if elapsed_time < self.AVOID_TURN_DURATION: self.publish_cmd(0.15, -0.8) # 右转避障 else: # 记录避障前的原始航向作为回正目标 self.target_yaw = self.current_yaw self.get_logger().info(f"📐 记录目标航向: {self.target_yaw:.2f} rad") self.state = 'realignment' self.avoid_start_time = now self.prev_error = 0.0 # 重置PID误差 self.integral = 0.0 # 重置积分项 self.get_logger().info("🔄 进入回正阶段") elif self.state == 'realignment': # 回正阶段（使用PID控制精准回正） angular_vel = self.pid_control(self.target_yaw, self.current_yaw) # 检查是否完成回正 error = abs(self.normalize_angle(self.target_yaw - self.current_yaw)) if error < self.YAW_THRESHOLD: self.state = 'normal' self.get_logger().info(f"✅ 回正完成, 误差: {error:.3f} rad") # 回正后尝试检测车道线 v, alpha, lane_detected = self.detect_lane_line(self.current_frame) if lane_detected: self.publish_cmd(v, alpha) else: self.publish_cmd(0.1, 0.0) # 低速直行寻找车道线 elif elapsed_time > self.REALIGN_TIMEOUT: self.state = 'normal' self.get_logger().warn("⏱️ 回正超时，强制恢复巡线") self.publish_cmd(0.1, 0.0) # 低速直行寻找车道线 else: # 使用PID控制进行回正 self.publish_cmd(0.1, angular_vel) self.get_logger().info(f"⏳ 回正中: 目标={self.target_yaw:.2f}, 当前={self.current_yaw:.2f}, 误差={error:.3f}") elif valid_obs and self.state == 'normal': # 检测到障碍物，开始避障 self.state = 'avoiding_turn' self.avoid_start_time = now self.get_logger().warn(f"🚧 检测到障碍物，距离={min(o[0] for o in valid_obs):.2f}m") else: # 正常巡线模式 v, alpha, lane_detected = self.detect_lane_line(self.current_frame) if lane_detected: self.publish_cmd(v, alpha) else: # 未检测到车道线时的安全策略 self.publish_cmd(0.05, self.last_valid_alpha) def publish_cmd(self, speed, steer): """发布控制指令""" msg = Twist() msg.linear.x = float(speed) msg.angular.z = float(steer) self.publisher_.publish(msg) def main(args=None): rclpy.init(args=args) node = LineAndObstacleNode() try: rclpy.spin(node) except KeyboardInterrupt: pass finally: node.destroy_node() rclpy.shutdown() if name == 'main': main() 按照这个来增加修改

C -->|存在| D[选择最优方向] C -->|不存在| E[旋转搜索] 数学表达： $$ \text{cost}(k) = \mu_1 \cdot \Delta(\theta_k,\theta_t) + \mu_2 \cdot \Delta(\theta_k,\theta_c) $$ 其中： - $\theta_t$：目标...

c++ 代码实现将下图代码的obj.seg渲染在image上 cv::Mat image = cv::imread("inference/yq.jpg"); auto yolo = yolo::load("/home/zhangyu/tensorrt_code/test/1.18-integrate-full-yolov5/workspace/last.transd.trtmodel", yolo::Type::V8Seg); if (yolo == nullptr) return; auto objs = yolo->forward(cvimg(image)); int i = 0; for (auto &obj : objs) { uint8_t b, g, r; tie(b, g, r) = yolo::random_color(obj.class_label); cv::rectangle(image, cv::Point(obj.left, obj.top), cv::Point(obj.right, obj.bottom), cv::Scalar(b, g, r), 5); auto name = cocolabels[obj.class_label]; auto caption = cv::format("%s %.2f", name, obj.confidence); int width = cv::getTextSize(caption, 0, 1, 2, nullptr).width + 10; cv::rectangle(image, cv::Point(obj.left - 3, obj.top - 33), cv::Point(obj.left + width, obj.top), cv::Scalar(b, g, r), -1); cv::putText(image, caption, cv::Point(obj.left, obj.top - 5), 0, 1, cv::Scalar::all(0), 2, 16); if (obj.seg) { // cv::imwrite(cv::format("%d_mask.jpg", i), // cv::Mat(obj.seg->height, obj.seg->width, CV_8U, obj.seg->data)); i++; } } printf("Save result to Result.jpg, %d objects\n", (int)objs.size()); cv::imwrite("Result.jpg", image); }

auto yolo = yolo::load("/home/zhangyu/tensorrt_code/test/1.18-integrate-full-yolov5/workspace/last.transd.trtmodel", yolo::Type::V8Seg); if (yolo == nullptr) return -1; // Perform object detection...

yolov5-6.0保存像素坐标

此段代码定义了一个名为save_detection_results()的新函数，它接收来自YOLOv5推断过程中的检测列表作为参数，并将其转化为易于理解的数据结构再导出为CSV表格形式。注意这里的路径设置应当根据实际情况调整。

import sys import cv2 import time import torch import traceback import threading import queue import dxcam import ctypes import os import glob import numpy as np import logitech.lg from PyQt6.QtWidgets import (QApplication, QMainWindow, QWidget, QVBoxLayout, QHBoxLayout, QLabel, QPushButton, QComboBox, QSlider, QSpinBox, QDoubleSpinBox, QLineEdit, QTabWidget, QGroupBox, QTextEdit, QFileDialog, QMessageBox, QSizePolicy, QSplitter, QDialog, QScrollArea) from PyQt6.QtCore import Qt, QTimer, QThread, pyqtSignal from PyQt6.QtGui import QImage, QPixmap, QPainter, QColor, QFont, QIcon, QKeyEvent, QMouseEvent from PyQt6.QtSvg import QSvgRenderer from PIL import Image from ultralytics import YOLO from pynput import mouse class PIDController: """PID控制器""" def init(self, kp, ki, kd, output_min=-100, output_max=100): self.kp = kp # 比例增益 self.ki = ki # 积分增益 self.kd = kd # 微分增益 self.output_min = output_min self.output_max = output_max # 状态变量 self.integral = 0.0 self.prev_error = 0.0 self.last_time = time.perf_counter() def compute(self, setpoint, current_value): """计算PID控制输出""" current_time = time.perf_counter() dt = current_time - self.last_time # 防止过小的时间差导致计算问题 MIN_DT = 0.0001 if dt < MIN_DT: dt = MIN_DT # 计算误差 error = setpoint - current_value # 比例项 P = self.kp * error # 积分项（防饱和） self.integral += error * dt I = self.ki * self.integral # 微分项 derivative = (error - self.prev_error) / dt D = self.kd * derivative # 合成输出 output = P + I + D # 输出限幅 if output > self.output_max: output = self.output_max elif output < self.output_min: output = self.output_min # 更新状态 self.prev_error = error self.last_time = current_time return output def reset(self): """重置控制器状态""" self.integral = 0.0 self.prev_error = 0.0 self.last_time = time.perf_counter() class ScreenDetector: def init(self, config_path): # 解析配置文件 self._parse_config(config_path) # 设备检测与模型加载 self.device = self._determine_device() self.model = YOLO(self.model_path).to(self.device) # 屏幕信息初始化 self._init_screen_info() # 控制参数初始化 self._init_control_params() # 状态管理 self.stop_event = threading.Event() self.camera_lock = threading.Lock() self.target_lock = threading.Lock() self.offset_lock = threading.Lock() self.button_lock = threading.Lock() # 推理状态控制 self.inference_active = False self.inference_lock = threading.Lock() # 初始化相机 self._init_camera() # 初始化鼠标监听器 self._init_mouse_listener() # 初始化PID控制器 self._init_pid_controllers() def _parse_config(self, config_path): """解析并存储配置参数""" self.cfg = self._parse_txt_config(config_path) # 存储常用参数 self.model_path = self.cfg['model_path'] self.model_device = self.cfg['model_device'] self.screen_target_size = int(self.cfg['screen_target_size']) self.detection_conf_thres = float(self.cfg['detection_conf_thres']) self.detection_iou_thres = float(self.cfg['detection_iou_thres']) self.detection_classes = [int(x) for x in self.cfg['detection_classes'].split(',')] self.visualization_color = tuple(map(int, self.cfg['visualization_color'].split(','))) self.visualization_line_width = int(self.cfg['visualization_line_width']) self.visualization_font_scale = float(self.cfg['visualization_font_scale']) self.visualization_show_conf = bool(self.cfg['visualization_show_conf']) self.fov_horizontal = float(self.cfg.get('move_fov_horizontal', '90')) self.mouse_dpi = int(self.cfg.get('move_mouse_dpi', '400')) self.target_offset_x_percent = float(self.cfg.get('target_offset_x', '50')) self.target_offset_y_percent = 100 - float(self.cfg.get('target_offset_y', '50')) # PID参数 self.pid_kp = float(self.cfg.get('pid_kp', '1.0')) self.pid_ki = float(self.cfg.get('pid_ki', '0.05')) self.pid_kd = float(self.cfg.get('pid_kd', '0.2')) # 贝塞尔曲线参数 self.bezier_steps = int(self.cfg.get('bezier_steps', '100')) self.bezier_duration = float(self.cfg.get('bezier_duration', '0.1')) self.bezier_curve = float(self.cfg.get('bezier_curve', '0.3')) def update_config(self, config_path): """动态更新配置""" try: # 重新解析配置文件 self._parse_config(config_path) # 更新可以直接修改的参数 self.detection_conf_thres = float(self.cfg['detection_conf_thres']) self.detection_iou_thres = float(self.cfg['detection_iou_thres']) self.target_offset_x_percent = float(self.cfg.get('target_offset_x', '50')) self.target_offset_y_percent = 100 - float(self.cfg.get('target_offset_y', '50')) # PID参数更新 self.pid_kp = float(self.cfg.get('pid_kp', '1.0')) self.pid_ki = float(self.cfg.get('pid_ki', '0.05')) self.pid_kd = float(self.cfg.get('pid_kd', '0.2')) # 更新PID控制器 self.pid_x = PIDController(self.pid_kp, self.pid_ki, self.pid_kd) self.pid_y = PIDController(self.pid_kp, self.pid_ki, self.pid_kd) # FOV和DPI更新 self.fov_horizontal = float(self.cfg.get('move_fov_horizontal', '90')) self.mouse_dpi = int(self.cfg.get('move_mouse_dpi', '400')) # 更新贝塞尔曲线参数 self.bezier_steps = int(self.cfg.get('bezier_steps', '100')) self.bezier_duration = float(self.cfg.get('bezier_duration', '0.1')) self.bezier_curve = float(self.cfg.get('bezier_curve', '0.3')) print("配置已动态更新") return True except Exception as e: print(f"更新配置失败: {str(e)}") traceback.print_exc() return False def _parse_txt_config(self, path): """解析TXT格式的配置文件""" config = {} with open(path, 'r', encoding='utf-8') as f: for line in f: line = line.strip() if not line or line.startswith('#'): continue if '=' in line: key, value = line.split('=', 1) config[key.strip()] = value.strip() return config def _init_pid_controllers(self): """初始化PID控制器""" # 创建XY方向的PID控制器 self.pid_x = PIDController(self.pid_kp, self.pid_ki, self.pid_kd) self.pid_y = PIDController(self.pid_kp, self.pid_ki, self.pid_kd) def start_inference(self): """启动推理""" with self.inference_lock: self.inference_active = True def stop_inference(self): """停止推理""" with self.inference_lock: self.inference_active = False def _determine_device(self): """确定运行设备""" if self.model_device == 'auto': return 'cuda' if torch.cuda.is_available() and torch.cuda.device_count() > 0 else 'cpu' return self.model_device def _init_screen_info(self): """初始化屏幕信息""" user32 = ctypes.windll.user32 self.screen_width, self.screen_height = user32.GetSystemMetrics(0), user32.GetSystemMetrics(1) self.screen_center = (self.screen_width // 2, self.screen_height // 2) # 计算截图区域 left = (self.screen_width - self.screen_target_size) // 2 top = (self.screen_height - self.screen_target_size) // 2 self.region = ( max(0, int(left)), max(0, int(top)), min(self.screen_width, int(left + self.screen_target_size)), min(self.screen_height, int(top + self.screen_target_size)) ) def _init_control_params(self): """初始化控制参数""" self.previous_target_info = None self.closest_target_absolute = None self.target_offset = None self.right_button_pressed = False # 改为鼠标右键状态 def _init_camera(self): """初始化相机""" try: with self.camera_lock: self.camera = dxcam.create( output_idx=0, output_color="BGR", region=self.region ) self.camera.start(target_fps=120, video_mode=True) except Exception as e: print(f"相机初始化失败: {str(e)}") try: # 降级模式 with self.camera_lock: self.camera = dxcam.create() self.camera.start(target_fps=60, video_mode=True) except Exception as fallback_e: print(f"降级模式初始化失败: {str(fallback_e)}") self.camera = None def _init_mouse_listener(self): """初始化鼠标监听器""" self.mouse_listener = mouse.Listener( on_click=self.on_mouse_click # 监听鼠标点击事件 ) self.mouse_listener.daemon = True self.mouse_listener.start() def on_mouse_click(self, x, y, button, pressed): """处理鼠标点击事件""" try: if button == mouse.Button.right: # 监听鼠标右键 with self.button_lock: self.right_button_pressed = pressed # 更新状态 # 当右键释放时重置PID控制器 if not pressed: self.pid_x.reset() self.pid_y.reset() except Exception as e: print(f"鼠标事件处理错误: {str(e)}") def calculate_fov_movement(self, dx, dy): """基于FOV算法计算鼠标移动量""" # 计算屏幕对角线长度 screen_diagonal = (self.screen_width 2 + self.screen_height 2) ** 0.5 # 计算垂直FOV aspect_ratio = self.screen_width / self.screen_height fov_vertical = self.fov_horizontal / aspect_ratio # 计算每像素对应角度 angle_per_pixel_x = self.fov_horizontal / self.screen_width angle_per_pixel_y = fov_vertical / self.screen_height # 计算角度偏移 angle_offset_x = dx * angle_per_pixel_x angle_offset_y = dy * angle_per_pixel_y # 转换为鼠标移动量 move_x = (angle_offset_x / 360) * self.mouse_dpi move_y = (angle_offset_y / 360) * self.mouse_dpi return move_x, move_y def move_mouse_to_target(self): """移动鼠标对准目标点""" if not self.target_offset: return try: # 获取目标点与屏幕中心的偏移量 with self.offset_lock: dx, dy = self.target_offset # 使用FOV算法将像素偏移转换为鼠标移动量 move_x, move_y = self.calculate_fov_movement(dx, dy) # 使用PID计算平滑的移动量 pid_move_x = self.pid_x.compute(0, -move_x) # 将dx取反 pid_move_y = self.pid_y.compute(0, -move_y) # 将dy取反 # 移动鼠标 if pid_move_x != 0 or pid_move_y != 0: logitech.lg.start_mouse_move(int(pid_move_x), int(pid_move_y), self.bezier_steps, self.bezier_duration, self.bezier_curve) except Exception as e: print(f"移动鼠标时出错: {str(e)}") def run(self, frame_queue): """主检测循环""" while not self.stop_event.is_set(): try: # 检查推理状态 with self.inference_lock: if not self.inference_active: time.sleep(0.01) continue # 截图 grab_start = time.perf_counter() screenshot = self._grab_screenshot() grab_time = (time.perf_counter() - grab_start) * 1000 # ms if screenshot is None: time.sleep(0.001) continue # 推理 inference_start = time.perf_counter() results = self._inference(screenshot) inference_time = (time.perf_counter() - inference_start) * 1000 # ms # 处理检测结果 target_info, closest_target_relative, closest_offset = self._process_detection_results(results) # 更新目标信息 self._update_target_info(target_info, closest_offset) # 移动鼠标 self._move_mouse_if_needed() # 可视化处理 annotated_frame = self._visualize_results(results, closest_target_relative) if frame_queue else None # 放入队列 if frame_queue: try: frame_queue.put( (annotated_frame, len(target_info), inference_time, grab_time, target_info), timeout=0.01 ) except queue.Full: pass except Exception as e: print(f"检测循环异常: {str(e)}") traceback.print_exc() self._reset_camera() time.sleep(0.5) def _grab_screenshot(self): """安全获取截图""" with self.camera_lock: if self.camera: return self.camera.grab() return None def _inference(self, screenshot): """执行模型推理""" return self.model.predict( screenshot, conf=self.detection_conf_thres, iou=self.detection_iou_thres, classes=self.detection_classes, device=self.device, verbose=False ) def _process_detection_results(self, results): """处理检测结果""" target_info = [] min_distance = float('inf') closest_target_relative = None closest_target_absolute = None closest_offset = None for box in results[0].boxes: # 获取边界框坐标 x1, y1, x2, y2 = map(int, box.xyxy[0]) # 计算绝对坐标 x1_abs = x1 + self.region[0] y1_abs = y1 + self.region[1] x2_abs = x2 + self.region[0] y2_abs = y2 + self.region[1] # 计算边界框尺寸 width = x2_abs - x1_abs height = y2_abs - y1_abs # 应用偏移百分比计算目标点 target_x = x1_abs + int(width * (self.target_offset_x_percent / 100)) target_y = y1_abs + int(height * (self.target_offset_y_percent / 100)) # 计算偏移量 dx = target_x - self.screen_center[0] dy = target_y - self.screen_center[1] distance = (dx 2 + dy 2) ** 0.5 # 更新最近目标 if distance < min_distance: min_distance = distance # 计算相对坐标（用于可视化） closest_target_relative = ( x1 + int(width * (self.target_offset_x_percent / 100)), y1 + int(height * (self.target_offset_y_percent / 100)) ) closest_target_absolute = (target_x, target_y) closest_offset = (dx, dy) # 保存目标信息 class_id = int(box.cls) class_name = self.model.names[class_id] target_info.append(f"{class_name}:{x1_abs},{y1_abs},{x2_abs},{y2_abs}") return target_info, closest_target_relative, closest_offset def _update_target_info(self, target_info, closest_offset): """更新目标信息""" # 检查目标信息是否有变化 if target_info != self.previous_target_info: self.previous_target_info = target_info.copy() print(f"{len(target_info)}|{'|'.join(target_info)}") # 更新目标偏移量 with self.offset_lock: self.target_offset = closest_offset def _visualize_results(self, results, closest_target): """可视化处理结果""" frame = results[0].plot( line_width=self.visualization_line_width, font_size=self.visualization_font_scale, conf=self.visualization_show_conf ) # 绘制最近目标 if closest_target: # 绘制目标中心点 cv2.circle( frame, (int(closest_target[0]), int(closest_target[1])), 3, (0, 0, 255), -1 ) # 计算屏幕中心在截图区域内的相对坐标 screen_center_x = self.screen_center[0] - self.region[0] screen_center_y = self.screen_center[1] - self.region[1] # 绘制中心到目标的连线 cv2.line( frame, (int(screen_center_x), int(screen_center_y)), (int(closest_target[0]), int(closest_target[1])), (0, 255, 0), 1 ) return frame def _move_mouse_if_needed(self): """如果需要则移动鼠标""" with self.button_lock: if self.right_button_pressed and self.target_offset: # 使用right_button_pressed self.move_mouse_to_target() def _reset_camera(self): """重置相机""" print("正在重置相机...") try: self._init_camera() except Exception as e: print(f"相机重置失败: {str(e)}") traceback.print_exc() def stop(self): """安全停止检测器""" self.stop_event.set() self._safe_stop() if hasattr(self, 'mouse_listener') and self.mouse_listener.running: # 改为停止鼠标监听器 self.mouse_listener.stop() def _safe_stop(self): """同步释放资源""" print("正在安全停止相机...") try: with self.camera_lock: if self.camera: self.camera.stop() print("相机已停止") except Exception as e: print(f"停止相机时发生错误: {str(e)}") print("屏幕检测器已停止") class DetectionThread(QThread): update_signal = pyqtSignal(object) def init(self, detector, frame_queue): super().init() self.detector = detector self.frame_queue = frame_queue self.running = True def run(self): self.detector.run(self.frame_queue) def stop(self): self.running = False self.detector.stop() class MainWindow(QMainWindow): def init(self, detector): super().init() self.detector = detector self.setWindowTitle("EFAI 1.1") self.setGeometry(100, 100, 600, 400) # 添加缺失的属性初始化 self.visualization_enabled = True self.inference_active = False # 初始推理状态为停止 #窗口置顶 self.setWindowFlag(Qt.WindowType.WindowStaysOnTopHint) # 创建帧队列 self.frame_queue = queue.Queue(maxsize=3) # 初始化UI self.init_ui() # 启动检测线程 self.detection_thread = DetectionThread(self.detector, self.frame_queue) self.detection_thread.start() # 启动UI更新定时器 self.update_timer = QTimer() self.update_timer.timeout.connect(self.update_ui) self.update_timer.start(1) # 每1ms更新一次 def toggle_visualization(self): # 实际更新可视化状态属性 self.visualization_enabled = not self.visualization_enabled # 更新按钮文本 if self.visualization_enabled: self.toggle_visualization_btn.setText("禁用可视化") else: self.toggle_visualization_btn.setText("启用可视化") def toggle_inference(self): """切换推理状态""" self.inference_active = not self.inference_active if self.inference_active: self.toggle_inference_btn.setText("停止推理") self.toggle_inference_btn.setStyleSheet(""" QPushButton { background-color: #F44336; color: white; border: none; padding: 8px; border-radius: 4px; font-family: Segoe UI; font-size: 10pt; } """) self.detector.start_inference() else: self.toggle_inference_btn.setText("开始推理") self.toggle_inference_btn.setStyleSheet(""" QPushButton { background-color: #4CAF50; color: white; border: none; padding: 8px; border-radius: 4px; font-family: Segoe UI; font-size: 10pt; } """) self.detector.stop_inference() def init_ui(self): # 主布局 central_widget = QWidget() self.setCentralWidget(central_widget) main_layout = QVBoxLayout(central_widget) # 分割器（左侧图像/目标信息，右侧控制面板） splitter = QSplitter(Qt.Orientation.Horizontal) main_layout.addWidget(splitter) # 左侧区域（图像显示和目标信息） left_widget = QWidget() left_layout = QVBoxLayout(left_widget) # 图像显示区域 self.image_label = QLabel() self.image_label.setAlignment(Qt.AlignmentFlag.AlignCenter) self.image_label.setMinimumSize(320, 320) left_layout.addWidget(self.image_label) # 目标信息区域 self.target_info_text = QTextEdit() self.target_info_text.setReadOnly(True) self.target_info_text.setFixedHeight(150) self.target_info_text.setStyleSheet(""" QTextEdit { background-color: #2D2D30; color: #DCDCDC; font-family: Consolas; font-size: 10pt; border: 1px solid #3F3F46; border-radius: 4px; } """) left_layout.addWidget(self.target_info_text) # 右侧控制面板 right_widget = QWidget() right_layout = QVBoxLayout(right_widget) right_layout.setAlignment(Qt.AlignmentFlag.AlignTop) # 性能信息 perf_group = QGroupBox("性能信息") perf_layout = QVBoxLayout(perf_group) self.target_count_label = QLabel("目标数量: 0") self.inference_time_label = QLabel("推理时间: 0.000s") self.grab_time_label = QLabel("截图时间: 0.000s") for label in [self.target_count_label, self.inference_time_label, self.grab_time_label]: label.setStyleSheet("font-family: Consolas; font-size: 10pt;") perf_layout.addWidget(label) right_layout.addWidget(perf_group) # 系统信息 sys_group = QGroupBox("系统信息") sys_layout = QVBoxLayout(sys_group) # 获取模型名称（只显示文件名） model_name = os.path.basename(self.detector.model_path) # 获取显示器编号（如果配置中有则显示，否则显示默认值0） monitor_index = self.detector.cfg.get('screen_monitor', '0') self.model_label = QLabel(f"模型: {model_name}") self.device_label = QLabel(f"设备: {self.detector.device.upper()}") self.monitor_label = QLabel(f"显示器:{monitor_index}") self.screen_res_label = QLabel(f"屏幕分辨率: {self.detector.screen_width}x{self.detector.screen_height}") self.region_label = QLabel(f"检测区域: {self.detector.region}") for label in [self.model_label, self.device_label, self.monitor_label, self.screen_res_label, self.region_label]: label.setStyleSheet("font-family: Consolas; font-size: 9pt; color: #A0A0A0;") sys_layout.addWidget(label) right_layout.addWidget(sys_group) # 鼠标状态 mouse_group = QGroupBox("自瞄状态") mouse_layout = QVBoxLayout(mouse_group) self.mouse_status = QLabel("未瞄准") self.mouse_status.setStyleSheet(""" QLabel { font-family: Consolas; font-size: 10pt; color: #FF5252; } """) mouse_layout.addWidget(self.mouse_status) right_layout.addWidget(mouse_group) # 控制按钮 btn_group = QGroupBox("控制") btn_layout = QVBoxLayout(btn_group) # 添加推理切换按钮 self.toggle_inference_btn = QPushButton("开始推理") self.toggle_inference_btn.clicked.connect(self.toggle_inference) self.toggle_inference_btn.setStyleSheet(""" QPushButton { background-color: #4CAF50; color: white; border: none; padding: 8px; border-radius: 4px; font-family: Segoe UI; font-size: 10pt; } QPushButton:hover { background-color: #45A049; } QPushButton:pressed { background-color: #3D8B40; } """) btn_layout.addWidget(self.toggle_inference_btn) self.toggle_visualization_btn = QPushButton("禁用可视化") self.toggle_visualization_btn.clicked.connect(self.toggle_visualization) self.settings_btn = QPushButton("设置") self.settings_btn.clicked.connect(self.open_settings) for btn in [self.toggle_visualization_btn, self.settings_btn]: btn.setStyleSheet(""" QPushButton { background-color: #0078D7; color: white; border: none; padding: 8px; border-radius: 4px; font-family: Segoe UI; font-size: 10pt; } QPushButton:hover { background-color: #106EBE; } QPushButton:pressed { background-color: #005A9E; } """) btn_layout.addWidget(btn) right_layout.addWidget(btn_group) # 添加左右区域到分割器 splitter.addWidget(left_widget) splitter.addWidget(right_widget) splitter.setSizes([600, 200]) # 设置样式 self.setStyleSheet(""" QMainWindow { background-color: #252526; } QGroupBox { font-family: Segoe UI; font-size: 10pt; color: #CCCCCC; border: 1px solid #3F3F46; border-radius: 4px; margin-top: 1ex; } QGroupBox::title { subcontrol-origin: margin; left: 10px; padding: 0 5px; background-color: transparent; } """) def open_settings(self): settings_dialog = SettingsDialog(self.detector.cfg, self) settings_dialog.exec() def update_ui(self): try: # 获取最新数据 latest_data = None while not self.frame_queue.empty(): latest_data = self.frame_queue.get_nowait() if latest_data: # 解包数据 frame, targets_count, inference_time, grab_time, target_info = latest_data # 更新性能信息 self.target_count_label.setText(f"目标数量: {targets_count}") self.inference_time_label.setText(f"推理时间: {inference_time / 1000:.3f}s") self.grab_time_label.setText(f"截图时间: {grab_time / 1000:.3f}s") # 更新目标信息 self.display_target_info(target_info) # 更新图像显示 if self.visualization_enabled and frame is not None: # 转换图像为Qt格式 height, width, channel = frame.shape bytes_per_line = 3 * width q_img = QImage(frame.data, width, height, bytes_per_line, QImage.Format.Format_BGR888) pixmap = QPixmap.fromImage(q_img) # 等比例缩放 scaled_pixmap = pixmap.scaled( self.image_label.width(), self.image_label.height(), Qt.AspectRatioMode.KeepAspectRatio, Qt.TransformationMode.SmoothTransformation ) self.image_label.setPixmap(scaled_pixmap) else: # 显示黑色背景 pixmap = QPixmap(self.image_label.size()) pixmap.fill(QColor(0, 0, 0)) self.image_label.setPixmap(pixmap) # 更新鼠标状态 self.update_mouse_status() except Exception as e: print(f"更新UI时出错: {str(e)}") def display_target_info(self, target_info): """在文本框中显示目标信息""" if not target_info: self.target_info_text.setPlainText("无检测目标") return info_text = "目标类别与坐标:\n" for i, data in enumerate(target_info): try: parts = data.split(":", 1) if len(parts) == 2: class_name, coords_str = parts coords = list(map(int, coords_str.split(','))) if len(coords) == 4: display_text = f"{class_name}: [{coords[0]}, {coords[1]}, {coords[2]}, {coords[3]}]" else: display_text = f"坐标格式错误: {data}" else: display_text = f"数据格式错误: {data}" except: display_text = f"解析错误: {data}" info_text += f"{display_text}\n" self.target_info_text.setPlainText(info_text) def update_mouse_status(self): """更新鼠标右键状态显示""" with self.detector.button_lock: if self.detector.right_button_pressed: self.mouse_status.setText("瞄准中") self.mouse_status.setStyleSheet("color: #4CAF50; font-family: Consolas; font-size: 10pt;") else: self.mouse_status.setText("未瞄准") self.mouse_status.setStyleSheet("color: #FF5252; font-family: Consolas; font-size: 10pt;") def closeEvent(self, event): """安全关闭程序""" self.detection_thread.stop() self.detection_thread.wait() event.accept() class SettingsDialog(QDialog): def init(self, config, parent=None): super().init(parent) self.config = config # 保存原始配置的副本用于比较 self.original_config = config.copy() self.setWindowTitle("设置") self.setGeometry(100, 100, 600, 500) self.init_ui() def init_ui(self): layout = QVBoxLayout() self.setLayout(layout) # 标签页 tabs = QTabWidget() layout.addWidget(tabs) # 检测设置标签页 detection_tab = QWidget() detection_layout = QVBoxLayout(detection_tab) self.create_detection_settings(detection_layout) tabs.addTab(detection_tab, "检测") # 移动设置标签页 move_tab = QWidget() move_layout = QVBoxLayout(move_tab) self.create_move_settings(move_layout) tabs.addTab(move_tab, "FOV") # 目标点设置标签页 target_tab = QWidget() target_layout = QVBoxLayout(target_tab) self.create_target_settings(target_layout) tabs.addTab(target_tab, "目标点") # PID设置标签页 pid_tab = QWidget() pid_layout = QVBoxLayout(pid_tab) self.create_pid_settings(pid_layout) tabs.addTab(pid_tab, "PID") # 贝塞尔曲线设置标签页 bezier_tab = QWidget() bezier_layout = QVBoxLayout(bezier_tab) self.create_bezier_settings(bezier_layout) tabs.addTab(bezier_tab, "贝塞尔曲线") # 按钮区域 btn_layout = QHBoxLayout() layout.addLayout(btn_layout) save_btn = QPushButton("保存配置") save_btn.clicked.connect(self.save_config) cancel_btn = QPushButton("取消") cancel_btn.clicked.connect(self.reject) for btn in [save_btn, cancel_btn]: btn.setStyleSheet(""" QPushButton { background-color: #0078D7; color: white; border: none; padding: 8px 16px; border-radius: 4px; font-family: Segoe UI; font-size: 10pt; } QPushButton:hover { background-color: #106EBE; } QPushButton:pressed { background-color: #005A9E; } """) btn_layout.addWidget(btn) btn_layout.addStretch() def create_detection_settings(self, layout): # 模型选择 model_group = QGroupBox("模型设置") model_layout = QVBoxLayout(model_group) # 获取基础路径 if getattr(sys, 'frozen', False): base_path = sys._MEIPASS else: base_path = os.path.dirname(os.path.abspath(file)) # 获取模型文件列表 models_dir = os.path.join(base_path, 'models') model_files = [] if os.path.exists(models_dir): model_files = glob.glob(os.path.join(models_dir, '.pt')) # 处理模型显示名称 model_display_names = [os.path.basename(f) for f in model_files] if model_files else ["未找到模型文件"] self.model_name_to_path = {os.path.basename(f): f for f in model_files} # 当前配置的模型处理 current_model_path = self.config['model_path'] current_model_name = os.path.basename(current_model_path) # 确保当前模型在列表中 if current_model_name not in model_display_names: model_display_names.append(current_model_name) self.model_name_to_path[current_model_name] = current_model_path # 模型选择下拉框 model_layout.addWidget(QLabel("选择模型:")) self.model_combo = QComboBox() self.model_combo.addItems(model_display_names) self.model_combo.setCurrentText(current_model_name) model_layout.addWidget(self.model_combo) # 设备选择 model_layout.addWidget(QLabel("运行设备:")) self.device_combo = QComboBox() self.device_combo.addItems(['auto', 'cuda', 'cpu']) self.device_combo.setCurrentText(self.config['model_device']) model_layout.addWidget(self.device_combo) layout.addWidget(model_group) # 检测参数 param_group = QGroupBox("检测参数") param_layout = QVBoxLayout(param_group) # 置信度阈值 param_layout.addWidget(QLabel("置信度阈值:")) conf_layout = QHBoxLayout() self.conf_slider = QSlider(Qt.Orientation.Horizontal) self.conf_slider.setRange(10, 100) # 0.1到1.0，步长0.01 self.conf_slider.setValue(int(float(self.config['detection_conf_thres']) 100)) conf_layout.addWidget(self.conf_slider) self.conf_value = QLabel(f"{float(self.config['detection_conf_thres']):.2f}") self.conf_value.setFixedWidth(50) conf_layout.addWidget(self.conf_value) param_layout.addLayout(conf_layout) # 连接滑块值变化事件 self.conf_slider.valueChanged.connect(lambda value: self.conf_value.setText(f"{value / 100:.2f}")) # IOU阈值 - 改为滑动条 param_layout.addWidget(QLabel("IOU阈值:")) iou_layout = QHBoxLayout() self.iou_slider = QSlider(Qt.Orientation.Horizontal) self.iou_slider.setRange(10, 100) # 0.1到1.0，步长0.01 self.iou_slider.setValue(int(float(self.config['detection_iou_thres']) * 100)) iou_layout.addWidget(self.iou_slider) self.iou_value = QLabel(f"{float(self.config['detection_iou_thres']):.2f}") self.iou_value.setFixedWidth(50) iou_layout.addWidget(self.iou_value) param_layout.addLayout(iou_layout) # 连接滑块值变化事件 self.iou_slider.valueChanged.connect(lambda value: self.iou_value.setText(f"{value / 100:.2f}")) # 检测类别 param_layout.addWidget(QLabel("检测类别 (逗号分隔):")) self.classes_edit = QLineEdit() self.classes_edit.setText(self.config['detection_classes']) param_layout.addWidget(self.classes_edit) layout.addWidget(param_group) # 屏幕设置 screen_group = QGroupBox("屏幕设置") screen_layout = QVBoxLayout(screen_group) # 显示器编号 screen_layout.addWidget(QLabel("显示器编号:")) self.monitor_spin = QSpinBox() self.monitor_spin.setRange(0, 3) # 假设最多支持4个显示器 self.monitor_spin.setValue(int(self.config.get('screen_monitor', '0'))) screen_layout.addWidget(self.monitor_spin) # 屏幕区域大小 screen_layout.addWidget(QLabel("截屏尺寸:")) self.screen_size_spin = QSpinBox() self.screen_size_spin.setRange(100, 2000) self.screen_size_spin.setValue(int(self.config['screen_target_size'])) screen_layout.addWidget(self.screen_size_spin) layout.addWidget(screen_group) layout.addStretch() def create_move_settings(self, layout): group = QGroupBox("鼠标移动参数") group_layout = QVBoxLayout(group) # FOV设置 group_layout.addWidget(QLabel("横向FOV(度):")) self.fov_spin = QDoubleSpinBox() self.fov_spin.setRange(1, 179) self.fov_spin.setValue(float(self.config.get('move_fov_horizontal', '90'))) group_layout.addWidget(self.fov_spin) # 鼠标DPI group_layout.addWidget(QLabel("鼠标DPI:")) self.dpi_spin = QSpinBox() self.dpi_spin.setRange(100, 20000) self.dpi_spin.setValue(int(self.config.get('move_mouse_dpi', '400'))) group_layout.addWidget(self.dpi_spin) layout.addWidget(group) layout.addStretch() def create_target_settings(self, layout): group = QGroupBox("目标点偏移") group_layout = QVBoxLayout(group) # X轴偏移 - 添加百分比显示 group_layout.addWidget(QLabel("X轴偏移:")) x_layout = QHBoxLayout() self.x_offset_slider = QSlider(Qt.Orientation.Horizontal) self.x_offset_slider.setRange(0, 100) self.x_offset_slider.setValue(int(float(self.config.get('target_offset_x', '50')))) x_layout.addWidget(self.x_offset_slider) self.x_offset_value = QLabel(f"{int(float(self.config.get('target_offset_x', '50')))}%") self.x_offset_value.setFixedWidth(50) x_layout.addWidget(self.x_offset_value) group_layout.addLayout(x_layout) # 连接滑块值变化事件 self.x_offset_slider.valueChanged.connect(lambda value: self.x_offset_value.setText(f"{value}%")) # Y轴偏移 - 添加百分比显示 group_layout.addWidget(QLabel("Y轴偏移:")) y_layout = QHBoxLayout() self.y_offset_slider = QSlider(Qt.Orientation.Horizontal) self.y_offset_slider.setRange(0, 100) self.y_offset_slider.setValue(int(float(self.config.get('target_offset_y', '50')))) y_layout.addWidget(self.y_offset_slider) self.y_offset_value = QLabel(f"{int(float(self.config.get('target_offset_y', '50')))}%") self.y_offset_value.setFixedWidth(50) y_layout.addWidget(self.y_offset_value) group_layout.addLayout(y_layout) # 连接滑块值变化事件 self.y_offset_slider.valueChanged.connect(lambda value: self.y_offset_value.setText(f"{value}%")) # 说明 info_label = QLabel("(0% = 左上角, 50% = 中心, 100% = 右下角)") info_label.setStyleSheet("font-size: 9pt; color: #888888;") group_layout.addWidget(info_label) layout.addWidget(group) layout.addStretch() def create_pid_settings(self, layout): group = QGroupBox("PID参数") group_layout = QVBoxLayout(group) # Kp参数 group_layout.addWidget(QLabel("比例增益(Kp):")) kp_layout = QHBoxLayout() self.kp_slider = QSlider(Qt.Orientation.Horizontal) self.kp_slider.setRange(1, 1000) # 0.01到10.0，步长0.01 self.kp_slider.setValue(int(float(self.config.get('pid_kp', '1.0')) * 100)) kp_layout.addWidget(self.kp_slider) self.kp_value = QLabel(f"{float(self.config.get('pid_kp', '1.0')):.2f}") self.kp_value.setFixedWidth(50) kp_layout.addWidget(self.kp_value) group_layout.addLayout(kp_layout) # 连接滑块值变化事件 self.kp_slider.valueChanged.connect(lambda value: self.kp_value.setText(f"{value / 100:.2f}")) # Ki参数 group_layout.addWidget(QLabel("积分增益(Ki):")) ki_layout = QHBoxLayout() self.ki_slider = QSlider(Qt.Orientation.Horizontal) self.ki_slider.setRange(0, 100) # 0.0000到0.1000，步长0.001 self.ki_slider.setValue(int(float(self.config.get('pid_ki', '0.05')) * 10000)) ki_layout.addWidget(self.ki_slider) self.ki_value = QLabel(f"{float(self.config.get('pid_ki', '0.05')):.4f}") self.ki_value.setFixedWidth(50) ki_layout.addWidget(self.ki_value) group_layout.addLayout(ki_layout) # 连接滑块值变化事件 self.ki_slider.valueChanged.connect(lambda value: self.ki_value.setText(f"{value / 10000:.4f}")) # Kd参数 group_layout.addWidget(QLabel("微分增益(Kd):")) kd_layout = QHBoxLayout() self.kd_slider = QSlider(Qt.Orientation.Horizontal) self.kd_slider.setRange(0, 5000) # 0.000到5.000，步长0.001 self.kd_slider.setValue(int(float(self.config.get('pid_kd', '0.2')) * 1000)) kd_layout.addWidget(self.kd_slider) self.kd_value = QLabel(f"{float(self.config.get('pid_kd', '0.2')):.3f}") self.kd_value.setFixedWidth(50) kd_layout.addWidget(self.kd_value) group_layout.addLayout(kd_layout) # 连接滑块值变化事件 self.kd_slider.valueChanged.connect(lambda value: self.kd_value.setText(f"{value / 1000:.3f}")) # 说明 info_text = "建议调整顺序: Kp → Kd → Ki\n\n" \ "先调整Kp至响应迅速但不过冲\n" \ "再增加Kd抑制震荡\n" \ "最后微调Ki消除剩余误差" info_label = QLabel(info_text) info_label.setStyleSheet("font-size: 9pt; color: #888888;") group_layout.addWidget(info_label) layout.addWidget(group) layout.addStretch() # 创建贝塞尔曲线设置 def create_bezier_settings(self, layout): group = QGroupBox("贝塞尔曲线参数") group_layout = QVBoxLayout(group) # 步数设置 group_layout.addWidget(QLabel("步数 (1-500):")) steps_layout = QHBoxLayout() self.steps_slider = QSlider(Qt.Orientation.Horizontal) self.steps_slider.setRange(1, 500) self.steps_slider.setValue(int(self.config.get('bezier_steps', 100))) steps_layout.addWidget(self.steps_slider) self.steps_value = QLabel(str(self.config.get('bezier_steps', 100))) self.steps_value.setFixedWidth(50) steps_layout.addWidget(self.steps_value) group_layout.addLayout(steps_layout) # 连接滑块值变化事件 self.steps_slider.valueChanged.connect(lambda value: self.steps_value.setText(str(value))) # 总移动时间设置 (秒) group_layout.addWidget(QLabel("总移动时间 (秒, 0.01-1.0):")) duration_layout = QHBoxLayout() self.duration_slider = QSlider(Qt.Orientation.Horizontal) self.duration_slider.setRange(1, 100) # 0.01到1.0，步长0.01 self.duration_slider.setValue(int(float(self.config.get('bezier_duration', 0.1)) * 100)) duration_layout.addWidget(self.duration_slider) self.duration_value = QLabel(f"{float(self.config.get('bezier_duration', 0.1)):.2f}") self.duration_value.setFixedWidth(50) duration_layout.addWidget(self.duration_value) group_layout.addLayout(duration_layout) # 连接滑块值变化事件 self.duration_slider.valueChanged.connect(lambda value: self.duration_value.setText(f"{value / 100:.2f}")) # 控制点偏移幅度 group_layout.addWidget(QLabel("控制点偏移幅度 (0-1):")) curve_layout = QHBoxLayout() self.curve_slider = QSlider(Qt.Orientation.Horizontal) self.curve_slider.setRange(0, 100) # 0.00到1.00，步长0.01 self.curve_slider.setValue(int(float(self.config.get('bezier_curve', 0.3)) * 100)) curve_layout.addWidget(self.curve_slider) self.curve_value = QLabel(f"{float(self.config.get('bezier_curve', 0.3)):.2f}") self.curve_value.setFixedWidth(50) curve_layout.addWidget(self.curve_value) group_layout.addLayout(curve_layout) # 连接滑块值变化事件 self.curve_slider.valueChanged.connect(lambda value: self.curve_value.setText(f"{value / 100:.2f}")) # 说明 info_text = "贝塞尔曲线参数说明:\n\n" \ "• 步数: 鼠标移动的细分步数，值越大移动越平滑\n" \ "• 总移动时间: 鼠标移动的总时间，值越小移动越快\n" \ "• 控制点偏移幅度: 控制贝塞尔曲线的弯曲程度，0为直线，1为最大弯曲" info_label = QLabel(info_text) info_label.setStyleSheet("font-size: 9pt; color: #888888;") group_layout.addWidget(info_label) layout.addWidget(group) layout.addStretch() def save_config(self): try: # 保存配置到字典 model_name = self.model_combo.currentText() model_path = self.model_name_to_path.get(model_name, model_name) self.config['model_path'] = model_path self.config['model_device'] = self.device_combo.currentText() self.config['screen_monitor'] = str(self.monitor_spin.value()) self.config['screen_target_size'] = str(self.screen_size_spin.value()) # 检测参数 self.config['detection_conf_thres'] = str(self.conf_slider.value() / 100) self.config['detection_iou_thres'] = str(self.iou_slider.value() / 100) self.config['detection_classes'] = self.classes_edit.text() # 移动设置 self.config['move_fov_horizontal'] = str(self.fov_spin.value()) self.config['move_mouse_dpi'] = str(self.dpi_spin.value()) # 目标点偏移设置 self.config['target_offset_x'] = str(self.x_offset_slider.value()) self.config['target_offset_y'] = str(self.y_offset_slider.value()) # PID设置 self.config['pid_kp'] = str(self.kp_slider.value() / 100) self.config['pid_ki'] = str(self.ki_slider.value() / 10000) self.config['pid_kd'] = str(self.kd_slider.value() / 1000) # 贝塞尔曲线设置 self.config['bezier_steps'] = str(self.steps_slider.value()) self.config['bezier_duration'] = str(self.duration_slider.value() / 100) self.config['bezier_curve'] = str(self.curve_slider.value() / 100) # 保存为TXT格式 with open('detection_config.txt', 'w', encoding='utf-8') as f: for key, value in self.config.items(): f.write(f"{key} = {value}\n") # 检查需要重启的参数是否被修改 restart_required = False restart_params = [] # 比较模型路径是否变化 if self.config['model_path'] != self.original_config.get('model_path', ''): restart_required = True restart_params.append("模型路径") # 比较设备类型是否变化 if self.config['model_device'] != self.original_config.get('model_device', ''): restart_required = True restart_params.append("设备类型") # 比较屏幕区域大小是否变化 if self.config['screen_target_size'] != self.original_config.get('screen_target_size', ''): restart_required = True restart_params.append("屏幕区域大小") # 比较检测类别是否变化 if self.config['detection_classes'] != self.original_config.get('detection_classes', ''): restart_required = True restart_params.append("检测类别") # 动态更新检测器配置 if self.parent() and hasattr(self.parent(), 'detector'): success = self.parent().detector.update_config('detection_config.txt') if success: if restart_required: # 需要重启的参数已修改 param_list = "、".join(restart_params) QMessageBox.information( self, "配置已保存", f"配置已保存！以下参数需要重启才能生效:\n{param_list}\n\n" "其他参数已实时更新。" ) else: # 所有参数都已实时更新 QMessageBox.information(self, "成功", "配置已实时更新生效！") else: QMessageBox.warning(self, "部分更新", "配置更新失败，请查看日志") else: QMessageBox.information(self, "成功", "配置已保存！部分参数需重启生效") self.accept() except Exception as e: QMessageBox.critical(self, "错误", f"保存配置失败: {str(e)}") if name == "main": detector = ScreenDetector('detection_config.txt') print(f"\nDXcam检测器初始化完成 | 设备: {detector.device.upper()}") app = QApplication(sys.argv) # 设置全局样式 app.setStyle("Fusion") app.setStyleSheet(""" QWidget { background-color: #252526; color: #D4D4D4; selection-background-color: #0078D7; selection-color: white; } QPushButton { background-color: #0078D7; color: white; border: none; padding: 5px 10px; border-radius: 4px; } QPushButton:hover { background-color: #106EBE; } QPushButton:pressed { background-color: #005A9E; } QComboBox, QLineEdit, QSpinBox, QDoubleSpinBox, QSlider { background-color: #3C3C40; color: #D4D4D4; border: 1px solid #3F3F46; border-radius: 4px; padding: 3px; } QComboBox:editable { background-color: #3C3C40; } QComboBox QAbstractItemView { background-color: #2D2D30; color: #D4D4D4; selection-background-color: #0078D7; selection-color: white; } QLabel { color: #D4D4D4; } QTabWidget::pane { border: 1px solid #3F3F46; background: #252526; } QTabBar::tab { background: #1E1E1E; color: #A0A0A0; padding: 8px 12px; border-top-left-radius: 4px; border-top-right-radius: 4px; } QTabBar::tab:selected { background: #252526; color: #FFFFFF; border-bottom: 2px solid #0078D7; } QTabBar::tab:hover { background: #2D2D30; } QGroupBox { background-color: #252526; border: 1px solid #3F3F46; border-radius: 4px; margin-top: 1ex; } QGroupBox::title { subcontrol-origin: margin; left: 10px; padding: 0 5px; background-color: transparent; color: #CCCCCC; } """) window = MainWindow(detector) window.show() sys.exit(app.exec()) 考虑所有问题，重构我的代码，最后给我完整的代码

output = P + I + D # 输出限幅 output = max(min(output, self.output_max), self.output_min) # 更新状态 self.prev_error = error self.last_time = current_time return output def reset(self)...

yolov5调用双目相机

### 集成和使用双目相机...此段脚本展示了如何从RealSense D435i这样的硬件采集RGB-D数据，并将其应用于改进后的YOLOv5实例化版本之中。注意这里仅提供了一个概念性的框架；具体实现细节可能因所使用的特定型号而异。

yolov5客流量监控

### 使用YOLOv5进行客流量监测 #### 客流量监测的重要性客流量监测对于商业场所管理和公共安全至关重要。通过精确统计进出特定区域的人数，管理人员可以获得有价值的数据支持决策过程[^2]。 #### 方案概述为了...

yolov5_d435i_detection-main

相关推荐

YOLOV3_Fire_Detection-yolo编程资源

部署_YOLOv5_detection_with_Streamlit_YOLOv5-Streamlit_Deploy

基于Intel_Realsense_D435i深度相机的YOLOv11实时目标检测与三维定位系统_使用PyTorch框架实现高效深度学习模型_支持80类COCO数据集目标识别_结合.zip

YOLOv5_DOTA_OBB-master：Windows环境中的性能优化技巧，提升部署效率

Yolov5与Intel Realsense D435i物体三维距离测量工具包

Handling Class Imbalance in YOLOv8 Object Detection Tasks

yolov5-6.0保存像素坐标

yolov5调用双目相机

yolov5客流量监控

大家在看

大唐杯仿真介绍.zip

桌面便签_SimpleStickyNotes.zip

美敦力BIS监护仪串口通讯协议手册

PFC与Fipy耦合技术：基于三角网格单元的双向流固耦合双轴压缩模拟,基于PFC流固耦合原理的双向耦合模拟技术：PFC与Fipy结合，三角网格单元实现渗流与双轴压缩模拟的双向交互作用 ,PFC流固耦合

KR C4 电气元器件检修服务 - 系统技术.pdf

最新推荐

[学习]电子支付与网络银行第六章电子支付法律法规.ppt

90%会计都不知道的Excel最常用九个快捷键!【2017-2018最新会计实务】.doc

Wamp5: 一键配置ASP/PHP/HTML服务器工具

【数据融合技术】：甘肃土壤类型空间分析中的专业性应用

sht20温湿度传感器使用什么将上拉电阻和滤波电容引出

Delphi仿速达财务软件导航条组件开发教程

【空间分布规律】：甘肃土壤类型与农业生产的关联性研究

常见运放电路的基本结构和基本原理

ASP.NET2.0初学者个人网站实例分享

【制图技术】：甘肃高质量土壤分布TIF图件的成图策略