yolov5 代码详解

目录

1. 目标检测
   • 数据集
   • 模型
   • 损失函数
   • 训练trian.py
   • 推理detec.py
2. 分类
   • 数据集
   • 模型
   • 损失函数
   • 训练trian.py
   • 推理detec.py
3. 分割
   • 数据集
   • 模型
   • 损失函数
   • 训练trian.py
   • 推理detec.py

YOLOV5是一款强大的模型，不仅包含分类、检测、分割（全能选手，谢谢大佬），而且在模型、数据增强、先验框和真实框的匹配、损失函数都有极大的改进。在推理速度和COCOmAP上也比之前的模型有提高。训练策略上也增加一些新的技巧，比如多尺度、rectangle、用遗传算法搜寻超参数。

1. 目标检测

1.1 数据集

1. 数据集存放格式

--datasets
    -- imags
        -- train
        -- val
        -- test
    -- labels
       -- train
       -- val
       -- test

在datasets文件夹分别存放imags和labels文件夹，imags文件夹存放图片信息(.jpg)。labels文件夹存放对应图片的标签信息（.txt）,标签信息包含目标物体的类别和真实框的坐标（cx,cy,w,h）,这些坐标都是归一化后的数据。data/coco128.yaml内存储数据地址和类别信息，用于训练需要。

2. 数据集处理流程
   加载数据集的函数

    train_loader, dataset = create_dataloader(train_path,
                                              imgsz,
                                              batch_size // WORLD_SIZE,
                                              gs,
                                              single_cls,
                                              hyp=hyp,
                                              augment=True,
                                              cache=None if opt.cache == 'val' else opt.cache,
                                              rect=opt.rect,
                                              rank=LOCAL_RANK,
                                              workers=workers,
                                              image_weights=opt.image_weights,
                                              quad=opt.quad,
                                              prefix=colorstr('train: '),
                                              shuffle=True,
                                              seed=opt.seed)
                                            

1.2 模型

1.3 损失函数

1.4 训练 trian.py

1.5 推理 detec.py

检测流程

1. 解析参数
2. 分析source类型
3. 模型运行结果保存路径
4. 加载模型
5. 加载数据
6. 推理：处理图片，预测，非极大值
7. 画框，保存结果

# YOLOv5  by Ultralytics, GPL-3.0 license
"""
AttributeError: partially initialized module 'cv2' has no attribute 'gapi_wip_gst_GStreamerPipeline' (most likely due to a circular import)

Run YOLOv5 detection inference on images, videos, directories, globs, YouTube, webcam, streams, etc.

Usage - sources:
    $ python detect.py --weights yolov5s.pt --source 0                               # webcam
                                                     img.jpg                         # image
                                                     vid.mp4                         # video
                                                     screen                          # screenshot
                                                     path/                           # directory
                                                     list.txt                        # list of images
                                                     list.streams                    # list of streams
                                                     'path/*.jpg'                    # glob
                                                     'https://youtu.be/Zgi9g1ksQHc'  # YouTube
                                                     'rtsp://example.com/media.mp4'  # RTSP, RTMP, HTTP stream

Usage - formats:
    $ python detect.py --weights yolov5s.pt                 # PyTorch
                                 yolov5s.torchscript        # TorchScript
                                 yolov5s.onnx               # ONNX Runtime or OpenCV DNN with --dnn
                                 yolov5s_openvino_model     # OpenVINO
                                 yolov5s.engine             # TensorRT
                                 yolov5s.mlmodel            # CoreML (macOS-only)
                                 yolov5s_saved_model        # TensorFlow SavedModel
                                 yolov5s.pb                 # TensorFlow GraphDef
                                 yolov5s.tflite             # TensorFlow Lite
                                 yolov5s_edgetpu.tflite     # TensorFlow Edge TPU
                                 yolov5s_paddle_model       # PaddlePaddle
"""

import argparse
import os
import platform
import sys
from pathlib import Path

import torch

FILE = Path(__file__).resolve()  #  '/Users/liushuang/Downloads/yolov5-master/detect.py'  当前文件路径
ROOT = FILE.parents[0]           #  '/Users/liushuang/Downloads/yolov5-master'  YOLOv5 root directory 当前文件路径的父目录
if str(ROOT) not in sys.path:    #  模块查询路径
    sys.path.append(str(ROOT))  # add ROOT to PATH
ROOT = Path(os.path.relpath(ROOT, Path.cwd()))  # relative   得到相对路径  '.'

from models.common import DetectMultiBackend
from utils.dataloaders import IMG_FORMATS, VID_FORMATS, LoadImages, LoadScreenshots, LoadStreams
from utils.general import (LOGGER, Profile, check_file, check_img_size, check_imshow, check_requirements, colorstr, cv2,
                           increment_path, non_max_suppression, print_args, scale_boxes, strip_optimizer, xyxy2xywh)
from utils.plots import Annotator, colors, save_one_box
from utils.torch_utils import select_device, smart_inference_mode


@smart_inference_mode()
def run(
        weights=ROOT / 'yolov5s.pt',  # model path or triton URL
        source=ROOT / 'data/images',  # file/dir/URL/glob/screen/0(webcam)
        data=ROOT / 'data/coco128.yaml',  # dataset.yaml path
        imgsz=(640, 640),  # inference size (height, width)
        conf_thres=0.25,  # confidence threshold
        iou_thres=0.45,  # NMS IOU threshold
        max_det=1000,  # maximum detections per image
        device='',  # cuda device, i.e. 0 or 0,1,2,3 or cpu
        view_img=False,  # show results
        save_txt=False,  # save results to *.txt
        save_conf=False,  # save confidences in --save-txt labels
        save_crop=False,  # save cropped prediction boxes
        nosave=False,  # do not save images/videos
        classes=None,  # filter by class: --class 0, or --class 0 2 3
        agnostic_nms=False,  # class-agnostic NMS
        augment=False,  # augmented inference
        visualize=False,  # visualize features
        update=False,  # update all models
        project=ROOT / 'runs/detect',  # save results to project/name
        name='exp',  # save results to project/name
        exist_ok=False,  # existing project/name ok, do not increment
        line_thickness=3,  # bounding box thickness (pixels)
        hide_labels=False,  # hide labels
        hide_conf=False,  # hide confidences
        half=False,  # use FP16 half-precision inference
        dnn=False,  # use OpenCV DNN for ONNX inference
        vid_stride=1,  # video frame-rate stride
):
    source = str(source)   #  'data/images'
    save_img = not nosave and not source.endswith('.txt')  # save inference images
    is_file = Path(source).suffix[1:] in (IMG_FORMATS + VID_FORMATS)  # suffix 后缀 '.jpg'   True  是视频或者图片？
    is_url = source.lower().startswith(('rtsp://', 'rtmp://', 'http://', 'https://'))   # 网址
    webcam = source.isnumeric() or source.endswith('.streams') or (is_url and not is_file)  # 摄像头
    screenshot = source.lower().startswith('screen')  # False
    if is_url and is_file:
        source = check_file(source)  # download

    # Directories
    save_dir = increment_path(Path(project) / name, exist_ok=exist_ok)  # increment run ； PosixPath('runs/detect/exp3')
    (save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True)  # make dir

    # Load model
    device = select_device(device)   # YOLOv5  2023-4-15 Python-3.10.10 torch-2.0.0 CPU
    model = DetectMultiBackend(weights, device=device, dnn=dnn, data=data, fp16=half)  # 选择模型后端框架 YOLOv5s summary: 213 layers, 7225885 parameters, 0 gradients, 16.4 GFLOPs
    stride, names, pt = model.stride, model.names, model.pt      # 32, cls_names, 模型是否是pytorch True
    imgsz = check_img_size(imgsz, s=stride)                      # check image size  [640, 640] 图片尺寸是否是32的倍数

    # Dataloader
    bs = 1  # batch_size
    if webcam:   #  摄像头
        view_img = check_imshow(warn=True)
        dataset = LoadStreams(source, img_size=imgsz, stride=stride, auto=pt, vid_stride=vid_stride)
        bs = len(dataset)
    elif screenshot:
        dataset = LoadScreenshots(source, img_size=imgsz, stride=stride, auto=pt)
    else:    # 文件夹
        dataset = LoadImages(source, img_size=imgsz, stride=stride, auto=pt, vid_stride=vid_stride)
    vid_path, vid_writer = [None] * bs, [None] * bs

    # Run inference
    model.warmup(imgsz=(1 if pt or model.triton else bs, 3, *imgsz))  # warmup  imgsz = (1, 3, 640, 640)
    seen, windows, dt = 0, [], (Profile(), Profile(), Profile())  # 0,[],(,, )
    for path, im, im0s, vid_cap, s in dataset:  # /Users/liushuang/Downloads/yolov5-master/data/images/bus.jpg;None; image 1/2 /Users/liushuang/Downloads/yolov5-master/data/images/bus.jpg:
        with dt[0]:   # 这个是干啥用的？
            im = torch.from_numpy(im).to(model.device)   # torch.Size([3, 384, 640])
            im = im.half() if model.fp16 else im.float()  # uint8 to fp16/32   float32-->float16
            im /= 255  # 0 - 255 to 0.0 - 1.0
            if len(im.shape) == 3:
                im = im[None]  # expand for batch dim   torch.Size([1, 3, 384, 640])

        # Inference
        with dt[1]:
            visualize = increment_path(save_dir / Path(path).stem, mkdir=True) if visualize else False    # False
            pred = model(im, augment=augment, visualize=visualize)  # pred[0].shape: torch.Size([1, 15120, 85])  ;  pred[1][0]:torch.Size([1, 3, 48, 80, 85]);;;pred[1][1]:torch.Size([1, 3, 24, 40, 85]);;;pred[1][0]:torch.Size([1, 3, 12, 20, 85]);;;

        # NMS
        with dt[2]:
            pred = non_max_suppression(pred, conf_thres, iou_thres, classes, agnostic_nms, max_det=max_det)  # torch.Size([4, 4(boxes)+1(conf)+1(cls)])

        # Second-stage classifier (optional)
        # pred = utils.general.apply_classifier(pred, classifier_model, im, im0s)

        # Process predictions
        for i, det in enumerate(pred):  # per image
            seen += 1   # 记录观测到的物体
            if webcam:  # batch_size >= 1
                p, im0, frame = path[i], im0s[i].copy(), dataset.count
                s += f'{i}: '
            else:
                p, im0, frame = path, im0s.copy(), getattr(dataset, 'frame', 0)  # path,(720, 1280, 3),0

            p = Path(p)  # to Path
            save_path = str(save_dir / p.name)  # im.jpg   #'runs/detect/exp3/zidane.jpg'
            txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}')  # im.txt  #  'runs/detect/exp3/labels/zidane'
            s += '%gx%g ' % im.shape[2:]  # print string   # s:image 2/2 /Users/liushuang/Downloads/yolov5-master/data/images/zidane.jpg  ; im.shape[2:] : '384x640 '
            gn = torch.tensor(im0.shape)[[1, 0, 1, 0]]  # normalization gain whwh   # (720, 1280, 3) -->  tensor([1280,  720, 1280,  720])
            imc = im0.copy() if save_crop else im0  # for save_crop
            annotator = Annotator(im0, line_width=line_thickness, example=str(names))  # 绘图工具
            if len(det):
                # Rescale boxes from img_size to im0 size  把检测到的框映射到原图上
                det[:, :4] = scale_boxes(im.shape[2:], det[:, :4], im0.shape).round()   # im.shape[2:]:torch.Size([384, 640]) ; im0.shape: (720, 1280, 3)

                # Print results
                for c in det[:, 5].unique():
                    n = (det[:, 5] == c).sum()  # detections per class
                    s += f"{n} {names[int(c)]}{'s' * (n > 1)}, "  # add to string

                # Write results  保存结果
                for *xyxy, conf, cls in reversed(det):
                    if save_txt:  # Write to file
                        xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist()  # normalized xywh   box在原图上的相对位置
                        line = (cls, *xywh, conf) if save_conf else (cls, *xywh)  # label format
                        with open(f'{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 image   画到原图上
                        c = int(cls)  # integer class
                        label = None if hide_labels else (names[c] if hide_conf else f'{names[c]} {conf:.2f}')  # 'person 0.88'
                        annotator.box_label(xyxy, label, color=colors(c, True))
                    if save_crop:
                        save_one_box(xyxy, imc, file=save_dir / 'crops' / names[c] / f'{p.stem}.jpg', BGR=True)  # (684, 416, 3)

            # Stream results
            im0 = annotator.result()
            if view_img:
                if platform.system() == 'Linux' and p not in windows:
                    windows.append(p)
                    cv2.namedWindow(str(p), cv2.WINDOW_NORMAL | cv2.WINDOW_KEEPRATIO)  # allow window resize (Linux)
                    cv2.resizeWindow(str(p), im0.shape[1], im0.shape[0])
                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 'stream'
                    if vid_path[i] != save_path:  # new video
                        vid_path[i] = save_path
                        if isinstance(vid_writer[i], cv2.VideoWriter):
                            vid_writer[i].release()  # release previous video writer
                        if vid_cap:  # video
                            fps = 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:  # stream
                            fps, w, h = 30, im0.shape[1], im0.shape[0]
                        save_path = str(Path(save_path).with_suffix('.mp4'))  # force *.mp4 suffix on results videos
                        vid_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}{'' if len(det) else '(no detections), '}{dt[1].dt * 1E3:.1f}ms")

    # Print results
    t = tuple(x.t / seen * 1E3 for x in dt)  # speeds per image  dt是检测耗时，seen记录检测的物体数量
    LOGGER.info(f'Speed: %.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(f"Results saved to {colorstr('bold', save_dir)}{s}")
    if update:
        strip_optimizer(weights[0])  # update model (to fix SourceChangeWarning)


def parse_opt():
    parser = argparse.ArgumentParser()
    parser.add_argument('--weights', nargs='+', type=str, default=ROOT / 'yolov5s.pt', help='model path or triton URL')   # 权重
    parser.add_argument('--source', type=str, default=ROOT / 'data/images', help='file/dir/URL/glob/screen/0(webcam)')    # 检测对象
    parser.add_argument('--data', type=str, default=ROOT / 'data/my_data.yaml', help='(optional) dataset.yaml path')      # 数据
    parser.add_argument('--imgsz', '--img', '--img-size', nargs='+', type=int, default=[640], help='inference size h,w')
    parser.add_argument('--conf-thres', type=float, default=0.25, help='confidence threshold')      # 置信度阈值
    parser.add_argument('--iou-thres', type=float, default=0.45, help='NMS IoU threshold')          # IOU阈值
    parser.add_argument('--max-det', type=int, default=1000, help='maximum detections per image')   # 最大检测数量/每张图片
    parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu')       # 运算设备
    parser.add_argument('--view-img', action='store_true', help='show results')                     # 展示预测结果
    parser.add_argument('--save-txt', action='store_true', help='save results to *.txt')            # 保存labels
    parser.add_argument('--save-conf', action='store_true', help='save confidences in --save-txt labels')
    parser.add_argument('--save-crop', action='store_true', help='save cropped prediction boxes')   # 抠图
    parser.add_argument('--nosave', action='store_true', help='do not save images/videos')
    parser.add_argument('--classes', nargs='+', type=int, help='filter by class: --classes 0, or --classes 0 2 3')  # 从所有类别中选去感兴趣的个别类
    parser.add_argument('--agnostic-nms', action='store_true', help='class-agnostic NMS')
    parser.add_argument('--augment', action='store_true', help='augmented inference')              # 数据增强
    parser.add_argument('--visualize', action='store_true', help='visualize features')
    parser.add_argument('--update', action='store_true', help='update all models')                  # 更新
    parser.add_argument('--project', default=ROOT / 'runs/detect', help='save results to project/name')       # 保存路径
    parser.add_argument('--name', default='exp', help='save results to project/name')                         # 保存路径的子文件夹
    parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment') # 新预测文件
    parser.add_argument('--line-thickness', default=3, type=int, help='bounding box thickness (pixels)')      # 线条宽度
    parser.add_argument('--hide-labels', default=False, action='store_true', help='hide labels')
    parser.add_argument('--hide-conf', default=False, action='store_true', help='hide confidences')
    parser.add_argument('--half', action='store_true', help='use FP16 half-precision inference')   # 半，精度
    parser.add_argument('--dnn', action='store_true', help='use OpenCV DNN for ONNX inference')    # 分布式训练
    parser.add_argument('--vid-stride', type=int, default=1, help='video frame-rate stride')       # 取祯间隔时长
    opt = parser.parse_args()
    opt.imgsz *= 2 if len(opt.imgsz) == 1 else 1  # expand  [640]-->[640,640]
    print_args(vars(opt))   # 打印所有的参数
    return opt


def main(opt):
    check_requirements(exclude=('tensorboard', 'thop'))    # 检测requirements.txt里面的包有没有成功安装。
    run(**vars(opt))


if __name__ == '__main__':
    opt = parse_opt()   # 解析命令行参数
    main(opt)

2. 分类

2.1 数据集

2.2 模型

2.3 损失函数

2.4 训练 trian.py

2.5 推理 detec.py

3. 分割

3.1 数据集

3.2 模型

3.3 损失函数

3.4 训练 trian.py

3.5 推理 detec.py