yolov5实现单目测距的代码

使用YOLOv5实现单目测距可以分为以下几个步骤：

1. 对相机进行标定，获取相机的内参和外参矩阵。
2. 使用YOLOv5检测出场景中的物体，并获取其在图像中的尺寸。
3. 根据相机内参和外参矩阵，将物体在图像中的位置转换为相机坐标系下的坐标。
4. 根据物体在相机坐标系下的坐标和相机的位置，计算出物体到相机的距离。

下面是使用YOLOv5实现单目测距的代码示例：

import cv2
import numpy as np
import torch
import time
from models.experimental import attempt_load
from utils.general import non_max_suppression
from utils.torch_utils import select_device

class YOLOv5Detector:
    def __init__(self, weights_path='yolov5s.pt', device='cpu'):
        self.device = select_device(device)
        self.model = attempt_load(weights_path, map_location=self.device)
        self.model.eval()

    def detect(self, img, conf_thres=0.5, iou_thres=0.5):
        img = torch.from_numpy(img.transpose(2, 0, 1)).float().div(255.0).unsqueeze(0).to(self.device)
        pred = self.model(img)[0]
        pred = non_max_suppression(pred, conf_thres=conf_thres, iou_thres=iou_thres)

        res = []
        for det in pred:
            if det is not None and len(det):
                det[:, :4] = det[:, :4].clamp(0, img.shape[3])
                det[:, 5] *= det[:, 4]
                for box in det:
                    x1, y1, x2, y2, conf, cls = box.tolist()
                    res.append([x1, y1, x2, y2, conf, int(cls)])
        return res

def calc_distance(K, R, T, width, height, bbox):
    # 将bbox中心点的坐标由图像坐标系转换为相机坐标系
    x_center = (bbox[0] + bbox[2]) / 2.0
    y_center = (bbox[1] + bbox[3]) / 2.0
    p2d = np.array([[x_center, y_center]], dtype=np.float32)
    p3d = cv2.undistortPoints(p2d, K, np.array([0, 0, 0, 0]), R, K)
    # 计算物体到相机的距离
    z = T[0, 2] / (p3d[0, 0] - T[0, 0])
    return z

if __name__ == '__main__':
    # 加载YOLOv5模型
    detector = YOLOv5Detector(weights_path='yolov5s.pt', device='cpu')

    # 加载相机标定文件
    with np.load('calibration.npz') as calib:
        K, D, R, T = [calib[i] for i in ('K', 'D', 'R', 'T')]

    # 打开摄像头
    cap = cv2.VideoCapture(0)

    while True:
        ret, frame = cap.read()
        if not ret:
            break

        # 检测物体并计算距离
        start_time = time.time()
        bbox_list = detector.detect(frame)
        for bbox in bbox_list:
            x1, y1, x2, y2, conf, cls = bbox
            width = x2 - x1
            height = y2 - y1
            distance = calc_distance(K, R, T, width, height, bbox)
            print('class: {}, distance: {:.2f}m'.format(cls, distance))
            cv2.rectangle(frame, (int(x1), int(y1)), (int(x2), int(y2)), (0, 255, 0), 2)
            cv2.putText(frame, 'class: {}, distance: {:.2f}m'.format(cls, distance), (int(x1), int(y1) - 10),
                        cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
        end_time = time.time()
        fps = 1.0 / (end_time - start_time)
        cv2.putText(frame, 'FPS: {:.2f}'.format(fps), (20, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 2)

        # 显示图像
        cv2.imshow('frame', frame)
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break

    cap.release()
    cv2.destroyAllWindows()

需要注意的是，这里的相机标定文件是使用OpenCV的calibrateCamera函数进行标定得到的。在实际应用中，需要根据实际情况进行标定并获取相机的内参和外参矩阵。