做团购网站需要什么资质,传奇霸业网页游戏开服,网站建设需要上税吗,网站建设资质要求文章目录 YOLOV5 双目相机实现三维测距#xff08;新版本#xff09;1. 项目流程2. 测距原理3. 操作步骤和代码解析4. 实时检测5. 训练6. 源码下载 YOLOV5 双目相机实现三维测距#xff08;新版本#xff09;
本文主要是对此篇文章做一些改进#xff0c;以及解释读者在… 文章目录 YOLOV5 双目相机实现三维测距新版本1. 项目流程2. 测距原理3. 操作步骤和代码解析4. 实时检测5. 训练6. 源码下载 YOLOV5 双目相机实现三维测距新版本
本文主要是对此篇文章做一些改进以及解释读者在复现过程中遇到的问题完整代码在文章末尾
1. 项目流程
YOLOv5检测目标并提取其中心像素点坐标双目相机经过系列操作将像素点坐标转为深度三维坐标根据三维坐标计算距离将深度信息画图显示
2. 测距原理
如果想了解双目测距原理请移步该文章 双目三维测距python
3. 操作步骤和代码解析
下载 yolov5 6.1版本源码 之前用的是5.0源码代码太旧出现了不少问题故更新了一下创建一个detect-01.py文件文件里部分代码解析如下
双目相机参数stereoconfig.py 双目相机标定误差越小越好我这里误差为0.1尽量使误差在0.2以下
import numpy as np
# 双目相机参数
class stereoCamera(object):def __init__(self):self.cam_matrix_left np.array([[1101.89299, 0, 1119.89634],[0, 1100.75252, 636.75282],[0, 0, 1]])self.cam_matrix_right np.array([[1091.11026, 0, 1117.16592],[0, 1090.53772, 633.28256],[0, 0, 1]])self.distortion_l np.array([[-0.08369, 0.05367, -0.00138, -0.0009, 0]])self.distortion_r np.array([[-0.09585, 0.07391, -0.00065, -0.00083, 0]])self.R np.array([[1.0000, -0.000603116945856524, 0.00377055351856816],[0.000608108737333211, 1.0000, -0.00132288199083992],[-0.00376975166958581, 0.00132516525298933, 1.0000]])self.T np.array([[-119.99423], [-0.22807], [0.18540]])self.baseline 119.99423 测距代码部分解析
这一部分我直接计算了目标检测框中心点的深度值把中心点的深度值当作了距离。你也可以写个循环计算平均值或者中位数把他们当作深度值
if (accel_frame % fps_set 0):t3 time.time() thread.join()points_3d thread.get_result()t4 time.time() a points_3d[int(y_0), int(x_0), 0] / 1000b points_3d[int(y_0), int(x_0), 1] / 1000c points_3d[int(y_0), int(x_0), 2] / 1000dis ((a**2b**2c**2)**0.5)主代码detect-01.py 加入了多线程处理加快处理速度
import argparse
import os
import sys
from pathlib import Pathimport cv2
import torch
import torch.backends.cudnn as cudnnFILE Path(__file__).resolve()
ROOT FILE.parents[0] # YOLOv5 root directory
if str(ROOT) not in sys.path:sys.path.append(str(ROOT)) # add ROOT to PATH
ROOT Path(os.path.relpath(ROOT, Path.cwd())) # relativefrom models.common import DetectMultiBackend
from utils.datasets import IMG_FORMATS, VID_FORMATS, LoadImages, LoadStreams
from utils.general import (LOGGER, check_file, check_img_size, check_imshow, check_requirements, colorstr,increment_path, non_max_suppression, print_args, scale_coords, strip_optimizer, xyxy2xywh)
from utils.plots import Annotator, colors, save_one_box
from utils.torch_utils import select_device, time_sync
from stereo.dianyuntu_yolo import preprocess, undistortion, getRectifyTransform, draw_line, rectifyImage, \stereoMatchSGBM
from stereo import stereoconfig_040_2
from stereo.stereo import stereo_threading, MyThreadtorch.no_grad()
def run(weightsROOT / yolov5s.pt, # model.pt path(s)sourceROOT / data/images, # file/dir/URL/glob, 0 for webcamdataROOT / data/coco128.yaml, # dataset.yaml pathimgsz(640, 640), # inference size (height, width)conf_thres0.25, # confidence thresholdiou_thres0.45, # NMS IOU thresholdmax_det1000, # maximum detections per imagedevice, # cuda device, i.e. 0 or 0,1,2,3 or cpuview_imgFalse, # show resultssave_txtFalse, # save results to *.txtsave_confFalse, # save confidences in --save-txt labelssave_cropFalse, # save cropped prediction boxesnosaveFalse, # do not save images/videosclassesNone, # filter by class: --class 0, or --class 0 2 3agnostic_nmsFalse, # class-agnostic NMSaugmentFalse, # augmented inferencevisualizeFalse, # visualize featuresupdateFalse, # update all modelsprojectROOT / runs/detect, # save results to project/namenameexp, # save results to project/nameexist_okFalse, # existing project/name ok, do not incrementline_thickness3, # bounding box thickness (pixels)hide_labelsFalse, # hide labelshide_confFalse, # hide confidenceshalfFalse, # use FP16 half-precision inferencednnFalse, # use OpenCV DNN for ONNX inference):source str(source)save_img not nosave and not source.endswith(.txt) # save inference imagesis_file Path(source).suffix[1:] in (IMG_FORMATS VID_FORMATS)is_url source.lower().startswith((rtsp://, rtmp://, http://, https://))webcam source.isnumeric() or source.endswith(.txt) or (is_url and not is_file)if is_url and is_file:source check_file(source) # download# Directoriessave_dir increment_path(Path(project) / name, exist_okexist_ok) # increment run(save_dir / labels if save_txt else save_dir).mkdir(parentsTrue, exist_okTrue) # make dir# Load modeldevice select_device(device)model DetectMultiBackend(weights, devicedevice, dnndnn, datadata)stride, names, pt, jit, onnx, engine model.stride, model.names, model.pt, model.jit, model.onnx, model.engineimgsz check_img_size(imgsz, sstride) # check image size# Halfhalf (pt or jit or onnx or engine) and device.type ! cpu # FP16 supported on limited backends with CUDAif pt or jit:model.model.half() if half else model.model.float()# Dataloaderif webcam:view_img check_imshow()cudnn.benchmark True # set True to speed up constant image size inferencedataset LoadStreams(source, img_sizeimgsz, stridestride, autopt)bs len(dataset) # batch_sizeelse:dataset LoadImages(source, img_sizeimgsz, stridestride, autopt)bs 1 # batch_sizevid_path, vid_writer [None] * bs, [None] * bs# Run inferencemodel.warmup(imgsz(1 if pt else bs, 3, *imgsz), halfhalf) # warmupdt, seen [0.0, 0.0, 0.0], 0config stereoconfig_040_2.stereoCamera()# 立体校正map1x, map1y, map2x, map2y, Q getRectifyTransform(720, 1280, config)for path, im, im0s, vid_cap, s in dataset:t1 time_sync()im torch.from_numpy(im).to(device)im im.half() if half else im.float() # uint8 to fp16/32im / 255 # 0 - 255 to 0.0 - 1.0if len(im.shape) 3:im im[None] # expand for batch dimt2 time_sync()dt[0] t2 - t1# Inferencevisualize increment_path(save_dir / Path(path).stem, mkdirTrue) if visualize else Falsepred model(im, augmentaugment, visualizevisualize)t3 time_sync()dt[1] t3 - t2# NMSpred non_max_suppression(pred, conf_thres, iou_thres, classes, agnostic_nms, max_detmax_det)dt[2] time_sync() - t3# Second-stage classifier (optional)# pred utils.general.apply_classifier(pred, classifier_model, im, im0s)# Process predictionsfor i, det in enumerate(pred): # per imageseen 1if webcam: # batch_size 1p, im0, frame path[i], im0s[i].copy(), dataset.counts f{i}: else:p, im0, frame path, im0s.copy(), getattr(dataset, frame, 0)
################################################ start ##############################################thread MyThread(stereo_threading, args(config, im0, map1x, map1y, map2x, map2y, Q))thread.start()p Path(p) # to Pathsave_path str(save_dir / p.name) # im.jpgtxt_path str(save_dir / labels / p.stem) ( if dataset.mode image else f_{frame}) # im.txts %gx%g % im.shape[2:] # print stringgn torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwhimc im0.copy() if save_crop else im0 # for save_cropannotator Annotator(im0, line_widthline_thickness, examplestr(names))if len(det):# Rescale boxes from img_size to im0 sizedet[:, :4] scale_coords(im.shape[2:], det[:, :4], im0.shape).round()# Print resultsfor c in det[:, -1].unique():n (det[:, -1] c).sum() # detections per classs f{n} {names[int(c)]}{s * (n 1)}, # add to string# Write resultsfor *xyxy, conf, cls in reversed(det):if (0 xyxy[2] 1280):if save_txt: # Write to filexywh (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywhline (cls, *xywh, conf) if save_conf else (cls, *xywh) # label formatwith open(txt_path .txt, a) as f:f.write((%g * len(line)).rstrip() % line \n)if save_img or save_crop or view_img: # Add bbox to imagex_center (xyxy[0] xyxy[2]) / 2y_center (xyxy[1] xyxy[3]) / 2x_0 int(x_center)y_0 int(y_center)if (0 x_0 1280):x1 xyxy[0]x2 xyxy[2]y1 xyxy[1]y2 xyxy[3]thread.join()points_3d thread.get_result()a points_3d[int(y_0), int(x_0), 0] / 1000b points_3d[int(y_0), int(x_0), 1] / 1000c points_3d[int(y_0), int(x_0), 2] / 1000distance ((a ** 2 b ** 2 c ** 2) ** 0.5)if (distance ! 0): ## Add bbox to imagec int(cls) # integer classlabel None if hide_labels else (names[c] if hide_conf else f{names[c]} {conf:.2f})annotator.box_label(xyxy, label, colorcolors(c, True))print(点 (%d, %d) 的 %s 距离左摄像头的相对距离为 %0.2f m % (x_center, y_center, label, distance))text_dis_avg dis:%0.2fm % distance# only put dis on framecv2.putText(im0, text_dis_avg, (int(x1 (x2 - x1) 5), int(y1 30)), cv2.FONT_ITALIC,1.2, (255, 255, 255), 3)if save_crop:save_one_box(xyxy, imc, filesave_dir / crops / names[c] / f{p.stem}.jpg, BGRTrue)# Stream resultsim0 annotator.result()if view_img:cv2.namedWindow(Webcam, cv2.WINDOW_NORMAL)cv2.resizeWindow(Webcam, 1280, 480)cv2.moveWindow(Webcam, 0, 100)cv2.imshow(Webcam, im0)cv2.waitKey(1)# cv2.imshow(str(p), im0)# cv2.waitKey(1) # 1 millisecond# Save results (image with detections)if save_img:if dataset.mode image:cv2.imwrite(save_path, im0)else: # video or streamif vid_path[i] ! save_path: # new videovid_path[i] save_pathif isinstance(vid_writer[i], cv2.VideoWriter):vid_writer[i].release() # release previous video writerif vid_cap: # videofps vid_cap.get(cv2.CAP_PROP_FPS)w int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))h int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))else: # streamfps, w, h 30, im0.shape[1], im0.shape[0]save_path str(Path(save_path).with_suffix(.mp4)) # force *.mp4 suffix on results videosvid_writer[i] cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*mp4v), fps, (w, h))vid_writer[i].write(im0)# Print time (inference-only)LOGGER.info(f{s}Done. ({t3 - t2:.3f}s))# Print resultst tuple(x / seen * 1E3 for x in dt) # speeds per imageLOGGER.info(fSpeed: %.1fms pre-process, %.1fms inference, %.1fms NMS per image at shape {(1, 3, *imgsz)} % t)if save_txt or save_img:s f\n{len(list(save_dir.glob(labels/*.txt)))} labels saved to {save_dir / labels} if save_txt else LOGGER.info(fResults saved to {colorstr(bold, save_dir)}{s})if update:strip_optimizer(weights) # update model (to fix SourceChangeWarning)def parse_opt():parser argparse.ArgumentParser()parser.add_argument(--weights, nargs, typestr, defaultROOT / yolov5s.pt, helpmodel path(s))parser.add_argument(--source, typestr, defaultROOT / data/images/a1.mp4, helpfile/dir/URL/glob, 0 for webcam)parser.add_argument(--data, typestr, defaultROOT / data/coco128.yaml, help(optional) dataset.yaml path)parser.add_argument(--imgsz, --img, --img-size, nargs, typeint, default[640], helpinference size h,w)parser.add_argument(--conf-thres, typefloat, default0.25, helpconfidence threshold)parser.add_argument(--iou-thres, typefloat, default0.45, helpNMS IoU threshold)parser.add_argument(--max-det, typeint, default1000, helpmaximum detections per image)parser.add_argument(--device, default, helpcuda device, i.e. 0 or 0,1,2,3 or cpu)parser.add_argument(--view-img, actionstore_true, helpshow results)parser.add_argument(--save-txt, actionstore_true, helpsave results to *.txt)parser.add_argument(--save-conf, actionstore_true, helpsave confidences in --save-txt labels)parser.add_argument(--save-crop, actionstore_true, helpsave cropped prediction boxes)parser.add_argument(--nosave, actionstore_true, helpdo not save images/videos)parser.add_argument(--classes, nargs, typeint, helpfilter by class: --classes 0, or --classes 0 2 3)parser.add_argument(--agnostic-nms, actionstore_true, helpclass-agnostic NMS)parser.add_argument(--augment, actionstore_true, helpaugmented inference)parser.add_argument(--visualize, actionstore_true, helpvisualize features)parser.add_argument(--update, actionstore_true, helpupdate all models)parser.add_argument(--project, defaultROOT / runs/detect, helpsave results to project/name)parser.add_argument(--name, defaultexp, helpsave results to project/name)parser.add_argument(--exist-ok, actionstore_true, helpexisting project/name ok, do not increment)parser.add_argument(--line-thickness, default3, typeint, helpbounding box thickness (pixels))parser.add_argument(--hide-labels, defaultFalse, actionstore_true, helphide labels)parser.add_argument(--hide-conf, defaultFalse, actionstore_true, helphide confidences)parser.add_argument(--half, actionstore_true, helpuse FP16 half-precision inference)parser.add_argument(--dnn, actionstore_true, helpuse OpenCV DNN for ONNX inference)opt parser.parse_args()opt.imgsz * 2 if len(opt.imgsz) 1 else 1 # expandprint_args(FILE.stem, opt)return optdef main(opt):check_requirements(exclude(tensorboard, thop))run(**vars(opt))if __name__ __main__:opt parse_opt()main(opt)执行detect-01.py检测结果如下
4. 实时检测
1如想要调用摄像头检测直接把detect-01.py里的
parser.add_argument(--source, typestr, defaultROOT / data/images/a1.mp4, helpfile/dir/URL/glob, 0 for webcam)改为
parser.add_argument(--source, typestr, defaultROOT / 0)2需要注意的是代码设置的是检测分辨率为2560x720大小的图或者视频直接调用摄像头摄像头分辨率不一定为2560x720因此需要设定一下打开摄像头默认分辨率
打开utils/dataset.py文件找到class LoadStreams:这个类 将
for i, s in enumerate(sources): # index, source# Start thread to read frames from video streamst f{i 1}/{n}: {s}... if youtube.com/ in s or youtu.be/ in s: # if source is YouTube videocheck_requirements((pafy, youtube_dl2020.12.2))import pafys pafy.new(s).getbest(preftypemp4).url # YouTube URLs eval(s) if s.isnumeric() else s # i.e. s 0 local webcamcap cv2.VideoCapture(s)assert cap.isOpened(), f{st}Failed to open {s}w int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))h int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))fps cap.get(cv2.CAP_PROP_FPS) # warning: may return 0 or nanself.frames[i] max(int(cap.get(cv2.CAP_PROP_FRAME_COUNT)), 0) or float(inf) # infinite stream fallbackself.fps[i] max((fps if math.isfinite(fps) else 0) % 100, 0) or 30 # 30 FPS fallback_, self.imgs[i] cap.read() # guarantee first frameself.threads[i] Thread(targetself.update, args([i, cap, s]), daemonTrue)LOGGER.info(f{st} Success ({self.frames[i]} frames {w}x{h} at {self.fps[i]:.2f} FPS))self.threads[i].start()
LOGGER.info() # newline改为
for i, s in enumerate(sources):# Start the thread to read frames from the video streamprint(f{i 1}/{n}: {s}... , end)cap cv2.VideoCapture(eval(s) if s.isnumeric() else s)####################################################################################################imageWidth 2560imageHeight 720cap.set(cv2.CAP_PROP_FRAME_WIDTH, imageWidth)cap.set(cv2.CAP_PROP_FRAME_HEIGHT, imageHeight)assert cap.isOpened(), fFailed to open {s}w int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))h int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))fps 24#cap.get(cv2.CAP_PROP_FPS) % 100_, self.imgs[i] cap.read() # guarantee first framethread Thread(targetself.update, args([i, cap]), daemonTrue)print(f success ({w}x{h} at {fps:.2f} FPS).)thread.start()print() # newline这样就设置好了
5. 训练
数据集采用YOLO格式目录如下
dataset|coco|images|train2017|1.jpg2.jpgval2017|11.jpg22.jpglabels|train2017|1.txt2.txtval2017|11.txt22.txt打开data/coco.yaml文件把里边的内容修改如下(这里训练两个类别)
# YOLOv5 by Ultralytics, GPL-3.0 license
# COCO128 dataset https://www.kaggle.com/ultralytics/coco128 (first 128 images from COCO train2017) by Ultralytics
# Example usage: python train.py --data coco128.yaml
# parent
# ├── yolov5
# └── datasets
# └── coco128 ← downloads here
# Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
path: ./dataset/coco # dataset root dir
train: images/train2017 # train images (relative to path) 128 images
val: images/train2017 # val images (relative to path) 128 images
test: # test images (optional)# Classes
nc: 2 # number of classes
names: [person, bicycle] # class names同时把训练用的model/yolov5s.yaml文件的类别改成与上边对应的类别数接下来运行train.py即可
6. 源码下载
下载链接https://download.csdn.net/download/qq_45077760/89136394
