训练完一个YOLOV5模型后，可以使用模型快速生成新图片数据的xml标注文件，下面是生成xml标注文件的具体脚本：
# YOLOv5 by Ultralytics, GPL-3.0 license
import os
import sys
from pathlib import Path
import torch
import torch.backends.cudnn as cudnn
FILE = 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())) # relative
from 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.torch_utils import select_device, time_sync
import xml.etree.ElementTree as et
import glob
@torch.no_grad()
def run(weights=ROOT / 'runs/train/exp/weights/best.pt', # model.pt path(s)
source='/home/images/source', # file/dir/URL/glob, 0 for webcam
imgsz=(640, 640), # inference size (height, width)
conf_thres=0.4, # confidence threshold
iou_thres=0.2, # NMS IOU threshold
max_det=1000, # maximum detections per image
device='0', # 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
):
source = str(source)
# Directories
save_dir = increment_path(Path(project) / name, exist_ok=exist_ok) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Load model
device = select_device(device)
model = DetectMultiBackend(weights, device=device, dnn=dnn)
stride, names, pt, jit, onnx, engine = model.stride, model.names, model.pt, model.jit, model.onnx, model.engine
imgsz = check_img_size(imgsz, s=stride) # check image size
# print(names)
# names=["car", "van", "bus", "truck", "other", 'left_line', 'right_line']
# print(names)
# Half
half &= (pt or jit or engine) and device.type != 'cpu' # half precision only supported by PyTorch on CUDA
if pt or jit:
model.model.half() if half else model.model.float()
# Run inference
model.warmup(imgsz=(1, 3, *imgsz), half=half) # warmup
dt, seen = [0.0, 0.0, 0.0], 0
index = 0
image_list = glob.glob(os.path.join(source, "*.jpg"))
for image_path in image_list:
dataset = LoadImages(image_path, img_size=imgsz, stride=stride, auto=pt)
bs = 1 # batch_size
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 fp16/32
im /= 255 # 0 - 255 to 0.0 - 1.0
if len(im.shape) == 3:
im = im[None] # expand for batch dim
t2 = time_sync()
dt[0] += t2 - t1
# Inference
visualize = increment_path(save_dir / Path(path).stem, mkdir=True) if visualize else False
pred = model(im, augment=augment, visualize=visualize)
t3 = time_sync()
dt[1] += t3 - t2
# NMS
pred = non_max_suppression(pred, conf_thres, iou_thres, classes, agnostic_nms, max_det=max_det)
dt[2] += time_sync() - t3
# Process predictions
for i, det in enumerate(pred): # per image
# 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)
p = Path(p) # to Path
root = et.Element("annotation")
folder = et.SubElement(root, "folder")
folder.text = "test"
filename = et.SubElement(root, "filename")
filename.text = p.name
# path = et.SubElement(root, "path")
# path.text = json_path.replace(".txt", ".jpg")
source = et.SubElement(root, "source")
database = et.SubElement(source, "database")
database.text = "Unknown"
size = et.SubElement(root, "size")
width = et.SubElement(size, "width")
width.text = str(im0s.shape[1])
height = et.SubElement(size, "height")
height.text = str(im0s.shape[0])
depth = et.SubElement(size, "depth")
depth.text = "3"
segmented = et.SubElement(root, "segmented")
segmented.text = "0"
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(im.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
c = int(cls) # integer class
# print(c)
# if c == 5 or c == 6:
# continue
label = None if hide_labels else (names[c] if hide_conf else f'{names[c]} {conf:.2f}')
object = et.SubElement(root, "object")
name = et.SubElement(object, "name")
name.text = names[c]
# print(names[c])
pose = et.SubElement(object, "pose")
pose.text = "Unspecified"
truncated = et.SubElement(object, "truncated")
truncated.text = "0"
difficult = et.SubElement(object, "difficult")
difficult.text = "0"
occluded = et.SubElement(object, "occluded")
occluded.text = "0"
bndbox = et.SubElement(object, "bndbox")
xmin = et.SubElement(bndbox, "xmin")
xmin.text = str(int(xyxy[0]))
ymin = et.SubElement(bndbox, "ymin")
ymin.text = str(int(xyxy[1]))
xmax = et.SubElement(bndbox, "xmax")
xmax.text = str(int(xyxy[2]))
ymax = et.SubElement(bndbox, "ymax")
ymax.text = str(int(xyxy[3]))
pretty_xml(root, ' ', '\n')
tree = et.ElementTree(root)
tree.write(image_path.replace(".jpg", ".xml"), encoding="utf-8")
index += 1
print(image_path, index)
def pretty_xml(element, indent, newline, level=0): # elemnt为传进来的Elment类，参数indent用于缩进，newline用于换行
if element: # 判断element是否有子元素
if (element.text is None) or element.text.isspace(): # 如果element的text没有内容
element.text = newline + indent * (level + 1)
else:
element.text = newline + indent * (level + 1) + element.text.strip() + newline + indent * (level + 1)
# else: # 此处两行如果把注释去掉，Element的text也会另起一行
# element.text = newline + indent * (level + 1) + element.text.strip() + newline + indent * level
temp = list(element) # 将element转成list
for subelement in temp:
if temp.index(subelement) < (len(temp) - 1): # 如果不是list的最后一个元素，说明下一个行是同级别元素的起始，缩进应一致
subelement.tail = newline + indent * (level + 1)
else: # 如果是list的最后一个元素， 说明下一行是母元素的结束，缩进应该少一个
subelement.tail = newline + indent * level
pretty_xml(subelement, indent, newline, level=level + 1) # 对子元素进行递归操作
if __name__ == "__main__":
run()
以上代码需要修改run函数中的：weights为yolov5模型路径，source为新图片数据所在文件夹，生成的xml也在source路径下。亲测可用！
