模型标注使用说明
1. 安装 labelme
pip install labelme
2. 运行 labelme
labelme
运行界面
使用方法就是打开图片或者图片目录
然后 创建多边形 鼠标点选自己要标注的内容
点击保存存会自动存储为json
数据集目录结构
官方目录结构
datasets
└── coco
├── annotations
│ └── instances_val2017.json
├── images
│ ├── train2017
│ └── val2017
├── labels
│ ├── train2017
│ └── val2017
├── LICENSE
├── README.txt
├── test-dev2017.txt
├── train2017.txt
└── val2017.txt
coco.yaml
# Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
# COCO 2017 dataset http://cocodataset.org by Microsoft
# Example usage: python train.py --data coco.yaml
# parent
# ├── yolov5
# └── datasets
# └── coco ← downloads here (20.1 GB)
# 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: ../datasets/coco # dataset root dir
train: train2017.txt # train images (relative to 'path') 118287 images
val: val2017.txt # val images (relative to 'path') 5000 images
test: test-dev2017.txt # 20288 of 40670 images, submit to https://competitions.codalab.org/competitions/20794
# Classes
names:
0: person
1: bicycle
2: car
3: motorcycle
4: airplane
5: bus
6: train
7: truck
8: boat
9: traffic light
10: fire hydrant
11: stop sign
12: parking meter
13: bench
14: bird
15: cat
16: dog
17: horse
18: sheep
19: cow
20: elephant
21: bear
22: zebra
23: giraffe
24: backpack
25: umbrella
26: handbag
27: tie
28: suitcase
29: frisbee
30: skis
31: snowboard
32: sports ball
33: kite
34: baseball bat
35: baseball glove
36: skateboard
37: surfboard
38: tennis racket
39: bottle
40: wine glass
41: cup
42: fork
43: knife
44: spoon
45: bowl
46: banana
47: apple
48: sandwich
49: orange
50: broccoli
51: carrot
52: hot dog
53: pizza
54: donut
55: cake
56: chair
57: couch
58: potted plant
59: bed
60: dining table
61: toilet
62: tv
63: laptop
64: mouse
65: remote
66: keyboard
67: cell phone
68: microwave
69: oven
70: toaster
71: sink
72: refrigerator
73: book
74: clock
75: vase
76: scissors
77: teddy bear
78: hair drier
79: toothbrush
train2017.txt
./images/train2017/000000109622.jpg
./images/train2017/000000160694.jpg
./images/train2017/000000308590.jpg
./images/train2017/000000327573.jpg
./images/train2017/000000062929.jpg
./images/train2017/000000512793.jpg
./images/train2017/000000371735.jpg
./images/train2017/000000148118.jpg
./images/train2017/000000309856.jpg
./images/train2017/000000141882.jpg
./images/train2017/000000318783.jpg
./images/train2017/000000337760.jpg
./images/train2017/000000298197.jpg
./images/train2017/000000042421.jpg
test-dev2017.txt
./images/test2017/000000000001.jpg
./images/test2017/000000000016.jpg
./images/test2017/000000000057.jpg
./images/test2017/000000000069.jpg
./images/test2017/000000000080.jpg
./images/test2017/000000000090.jpg
./images/test2017/000000000106.jpg
./images/test2017/000000000108.jpg
./images/test2017/000000000128.jpg
./images/test2017/000000000155.jpg
./images/test2017/000000000171.jpg
./images/test2017/000000000178.jpg
./images/test2017/000000000180.jpg
./images/test2017/000000000188.jpg
./images/test2017/000000000202.jpg
./images/test2017/000000000229.jpg
./images/test2017/000000000275.jpg
./images/test2017/000000000276.jpg
./images/test2017/000000000318.jpg
./images/test2017/000000000408.jpg
./images/test2017/000000000427.jpg
./images/test2017/000000000448.jpg
./images/test2017/000000000453.jpg
./images/test2017/000000000456.jpg
./images/test2017/000000000463.jpg
./images/test2017/000000000492.jpg
val2017.txt
./images/val2017/000000182611.jpg
./images/val2017/000000335177.jpg
./images/val2017/000000278705.jpg
./images/val2017/000000463618.jpg
./images/val2017/000000568981.jpg
./images/val2017/000000092416.jpg
./images/val2017/000000173830.jpg
./images/val2017/000000476215.jpg
./images/val2017/000000479126.jpg
./images/val2017/000000570664.jpg
./images/val2017/000000304396.jpg
./images/val2017/000000231339.jpg
./images/val2017/000000153510.jpg
./images/val2017/000000011051.jpg
./images/val2017/000000237984.jpg
./images/val2017/000000124798.jpg
./images/val2017/000000441491.jpg
./images/val2017/000000361268.jpg
./images/val2017/000000272566.jpg
./images/val2017/000000160864.jpg
./images/val2017/000000078959.jpg
./images/val2017/000000047571.jpg
./images/val2017/000000311002.jpg
./images/val2017/000000328683.jpg
./images/val2017/000000289059.jpg
./images/val2017/000000559547.jpg
./images/val2017/000000577862.jpg
训练
使用官方数据训练
python train.py --img 640 --batch 16 --epochs 50 --data coco.yaml --weights yolov5s.pt
使用自定义数据训练
python train.py --img 640 --batch 48 --epochs 1000 --data ./datasets/dataset.yaml --weights yolov5s.pt --device 0
参数说明：
• --weights yolov5s.pt：使用 yolov5s 预训练模型。
• --source data/images：检测 data/images 文件夹中的图片，可以是单张图片、视频、摄像头（0）。
• --imgsz 640：输入图片大小，YOLOv5 默认 640x640。
• --conf-thres 0.25：置信度阈值（默认 0.25）。
• --iou-thres 0.45：IoU 阈值（默认 0.45）。
• --device 0：使用 GPU（0）或 CPU（cpu）。
• --save-txt：保存检测结果（文本）。
• --save-crop：保存裁剪出的目标图像。
| 模型 | 速度 | 精度 | 参数量 | FLOPs |
|---|---|---|---|---|
| yolov5n.pt | 🚀 超快 | 📉 低 | 1.9M | 4.5B |
| yolov5s.pt | ⚡ 快 | ⭐ 标准 | 7.2M | 16.5B |
| yolov5m.pt | 🔄 中 | 📈 高 | 21.2M | 49.0B |
| yolov5l.pt | 🐢 慢 | 📊 更高 | 46.5M | 115.4B |
| yolov5x.pt | 🏗 超慢 | 🔥 最高 | 86.7M | 205.7B |
下载模型
wget https://github.com/ultralytics/yolov5/releases/download/v7.0/yolov5n.pt
wget https://github.com/ultralytics/yolov5/releases/download/v7.0/yolov5m.pt
wget https://github.com/ultralytics/yolov5/releases/download/v7.0/yolov5l.pt
这段信息是 YOLOv5 训练过程中的一部分，具体的字段含义如下：
8/127：这是当前训练 epoch（训练轮次）和总轮次的进度，表示当前是第 8 个 epoch，总共有 127 个 epoch。
18.7G：当前模型训练使用的 GPU 显存量，18.7GB。
0.04157：当前训练批次的损失值（loss）。
0.06278 和 0.01883：通常表示其他的训练统计数据，如目标检测中的 IOU（Intersection over Union）等指标，具体取决于模型和任务。
204：当前训练批次的样本数（batch size）。
640：每张图像的分辨率是 640x640。
进度条：45%|████▌ 表示训练进度的可视化，已经完成了 45%。
3354/7393：表示当前训练处理到了第 3354 个批次，总共有 7393 个批次。
[11:08<13:23, 5.03it/s]：这部分表示：
当前时间是 11:08。
剩余的训练时间是 13 分 23 秒。
当前训练速度是每秒 5.03 个批次。
使用转换脚本将labelme标注的json文件转换成yolo格式
import os
import json
import cv2
import numpy as np
def convert_labelme_to_yolo(json_path, output_dir, class_names, image_dir):
with open(json_path, 'r', encoding='utf-8') as f:
data = json.load(f)
# 读取 JSON 内的 imagePath
image_path = data['imagePath']
# 确保使用完整路径
image_path = os.path.join(image_dir, os.path.basename(image_path))
if not os.path.exists(image_path):
print(f"Error: 图片 {image_path} 不存在")
return
image = cv2.imread(image_path)
if image is None:
print(f"Error: 读取图片失败 {image_path}")
return
h, w, _ = image.shape
# 生成 YOLO 标注文件
yolo_txt_path = os.path.join(output_dir, os.path.basename(json_path).replace('.json', '.txt'))
with open(yolo_txt_path, 'w') as f:
for shape in data['shapes']:
label = shape['label']
points = np.array(shape['points'])
x_min, y_min = points.min(axis=0)
x_max, y_max = points.max(axis=0)
x_center = (x_min + x_max) / 2 / w
y_center = (y_min + y_max) / 2 / h
width = (x_max - x_min) / w
height = (y_max - y_min) / h
class_id = class_names.index(label)
f.write(f"{class_id} {x_center} {y_center} {width} {height}\n")
# 你的数据路径
json_dir = r"C:\Users\lq\Desktop\datasets\json" # 这里是labelme 保存json的文件夹
image_dir = r"C:\Users\lq\Desktop\datasets\images\train" # 图片存放的文件夹
output_dir = r"C:\Users\lq\Desktop\datasets\labels\train" # 输出 YOLO 标注文件的文件夹
os.makedirs(output_dir, exist_ok=True)
# 设置类别 这里的类别要和你自己标注的一致
class_names = ["手机","手"]
# 处理所有 JSON 文件
for file in os.listdir(json_dir):
if file.endswith(".json"):
convert_labelme_to_yolo(os.path.join(json_dir, file), output_dir, class_names, image_dir)
yolo格式转 labelme 数据
import os
import json
import base64
import cv2
dataset_path = "coco"
image_folder = os.path.join(dataset_path, "images/train2017") # 图片路径
label_folder = os.path.join(dataset_path, "labels/train2017") # 标签路径
train_txt = os.path.join(dataset_path, "train2017.txt") # 训练集索引
output_json_folder = os.path.join(dataset_path, "labelme_jsons")# LabelMe JSON 输出目录
os.makedirs(output_json_folder, exist_ok=True)
LABELME_VERSION = "5.6.1"
CLASS_NAMES = {
0: "person", 1: "bicycle", 2: "car", 3: "motorcycle", 4: "airplane", 5: "bus", 6: "train", 7: "truck",
8: "boat", 9: "traffic light", 10: "fire hydrant", 11: "stop sign", 12: "parking meter", 13: "bench",
14: "bird", 15: "cat", 16: "dog", 17: "horse", 18: "sheep", 19: "cow", 20: "elephant", 21: "bear",
22: "zebra", 23: "giraffe", 24: "backpack", 25: "umbrella", 26: "handbag", 27: "tie", 28: "suitcase",
29: "frisbee", 30: "skis", 31: "snowboard", 32: "sports ball", 33: "kite", 34: "baseball bat",
35: "baseball glove", 36: "skateboard", 37: "surfboard", 38: "tennis racket", 39: "bottle",
40: "wine glass", 41: "cup", 42: "fork", 43: "knife", 44: "spoon", 45: "bowl", 46: "banana",
47: "apple", 48: "sandwich", 49: "orange", 50: "broccoli", 51: "carrot", 52: "hot dog", 53: "pizza",
54: "donut", 55: "cake", 56: "chair", 57: "couch", 58: "potted plant", 59: "bed", 60: "dining table",
61: "toilet", 62: "tv", 63: "laptop", 64: "mouse", 65: "remote", 66: "keyboard", 67: "cell phone",
68: "microwave", 69: "oven", 70: "toaster", 71: "sink", 72: "refrigerator", 73: "book", 74: "clock",
75: "vase", 76: "scissors", 77: "teddy bear", 78: "hair drier", 79: "toothbrush"
}
# -------------------------------------------------------------------------
def convert_yolo_to_labelme_polygon(image_path, label_path, output_json):
"""
将一张图片的 YOLO 标注 (txt) 转换成 LabelMe JSON (polygon 四点).
包含 base64 的 imageData 字段。
"""
img = cv2.imread(image_path)
if img is None:
print(f"无法读取图像: {image_path}")
return
h, w = img.shape[:2]
with open(image_path, "rb") as f:
image_data_bytes = f.read()
image_data_b64 = base64.b64encode(image_data_bytes).decode("utf-8")
shapes = []
if os.path.exists(label_path):
with open(label_path, "r") as f:
for line in f:
line = line.strip()
if not line:
continue
parts = line.split()
if len(parts) != 5:
continue
class_id, x_center, y_center, width, height = map(float, parts)
class_id = int(class_id)
x_center *= w
y_center *= h
width *= w
height *= h
x1 = x_center - width / 2
y1 = y_center - height / 2
x2 = x_center + width / 2
y2 = y_center + height / 2
points = [
[x1, y1],
[x2, y1],
[x2, y2],
[x1, y2]
]
label_name = CLASS_NAMES.get(class_id, f"class_{class_id}")
shape_item = {
"label": label_name,
"points": points,
"group_id": None,
"description": "",
"shape_type": "polygon",
"flags": {},
"mask": None
}
shapes.append(shape_item)
labelme_data = {
"version": LABELME_VERSION,
"flags": {},
"shapes": shapes,
"imagePath": os.path.basename(image_path),
"imageData": image_data_b64,
"imageHeight": h,
"imageWidth": w
}
with open(output_json, "w", encoding="utf-8") as f:
json.dump(labelme_data, f, indent=4, ensure_ascii=False)
print(f"转换完成: {output_json}")
def main():
with open(train_txt, "r", encoding="utf-8") as f:
lines = [l.strip() for l in f if l.strip()]
for img_rel_path in lines:
img_path = os.path.join(dataset_path, img_rel_path)
filename = os.path.basename(img_path)
name_only, _ = os.path.splitext(filename)
label_name = name_only + ".txt"
label_path = os.path.join(label_folder, label_name)
output_json_path = os.path.join(output_json_folder, name_only + ".json")
convert_yolo_to_labelme_polygon(img_path, label_path, output_json_path)
if __name__ == "__main__":
main()
import os
import json
import base64
import cv2
import numpy as np
dataset_path = "coco"
image_folder = os.path.join(dataset_path, "images/train2017") # 图片路径
label_folder = os.path.join(dataset_path, "labels/train2017") # 标签路径
train_txt = os.path.join(dataset_path, "train2017.txt") # 训练集索引
output_json_folder = os.path.join(dataset_path, "labelme_jsons")# LabelMe JSON 输出目录
os.makedirs(output_json_folder, exist_ok=True)
LABELME_VERSION = "5.6.1"
CLASS_NAMES = {
0: "person", 1: "bicycle", 2: "car", 3: "motorcycle", 4: "airplane", 5: "bus", 6: "train", 7: "truck",
8: "boat", 9: "traffic light", 10: "fire hydrant", 11: "stop sign", 12: "parking meter", 13: "bench",
14: "bird", 15: "cat", 16: "dog", 17: "horse", 18: "sheep", 19: "cow", 20: "elephant", 21: "bear",
22: "zebra", 23: "giraffe", 24: "backpack", 25: "umbrella", 26: "handbag", 27: "tie", 28: "suitcase",
29: "frisbee", 30: "skis", 31: "snowboard", 32: "sports ball", 33: "kite", 34: "baseball bat",
35: "baseball glove", 36: "skateboard", 37: "surfboard", 38: "tennis racket", 39: "bottle",
40: "wine glass", 41: "cup", 42: "fork", 43: "knife", 44: "spoon", 45: "bowl", 46: "banana",
47: "apple", 48: "sandwich", 49: "orange", 50: "broccoli", 51: "carrot", 52: "hot dog", 53: "pizza",
54: "donut", 55: "cake", 56: "chair", 57: "couch", 58: "potted plant", 59: "bed", 60: "dining table",
61: "toilet", 62: "tv", 63: "laptop", 64: "mouse", 65: "remote", 66: "keyboard", 67: "cell phone",
68: "microwave", 69: "oven", 70: "toaster", 71: "sink", 72: "refrigerator", 73: "book", 74: "clock",
75: "vase", 76: "scissors", 77: "teddy bear", 78: "hair drier", 79: "toothbrush"
}
# -------------------------------------------------------------------------
def convert_yolo_to_labelme_polygon(image_path, label_path, output_json):
"""
将一张图片的 YOLO 标注 (txt) 转换成 LabelMe JSON (polygon 多边形).
包含 base64 的 imageData 字段。
"""
def calculate_polygon(x_center, y_center, width, height, angle=0):
"""
计算一个矩形的四个角的坐标，并生成一个多边形。
参数 angle 可选，用于旋转矩形，单位为度数。
"""
w, h = width, height
# 计算矩形的四个角
box = np.array([
[-w / 2, -h / 2], # 左上
[w / 2, -h / 2], # 右上
[w / 2, h / 2], # 右下
[-w / 2, h / 2] # 左下
])
# 如果有旋转角度，进行旋转
if angle != 0:
angle_rad = np.deg2rad(angle)
rotation_matrix = np.array([
[np.cos(angle_rad), -np.sin(angle_rad)],
[np.sin(angle_rad), np.cos(angle_rad)]
])
box = np.dot(box, rotation_matrix)
# 平移到目标位置
box[:, 0] += x_center
box[:, 1] += y_center
return box
img = cv2.imread(image_path)
if img is None:
print(f"无法读取图像: {image_path}")
return
h, w = img.shape[:2]
with open(image_path, "rb") as f:
image_data_bytes = f.read()
image_data_b64 = base64.b64encode(image_data_bytes).decode("utf-8")
shapes = []
if os.path.exists(label_path):
with open(label_path, "r") as f:
for line in f:
line = line.strip()
if not line:
continue
parts = line.split()
if len(parts) != 5:
continue
class_id, x_center, y_center, width, height = map(float, parts)
class_id = int(class_id)
x_center *= w
y_center *= h
width *= w
height *= h
# 计算多边形的四个点
polygon_points = calculate_polygon(x_center, y_center, width, height)
label_name = CLASS_NAMES.get(class_id, f"class_{class_id}")
shape_item = {
"label": label_name,
"points": polygon_points.tolist(), # 转换为列表格式
"group_id": None,
"description": "",
"shape_type": "polygon",
"flags": {},
"mask": None
}
shapes.append(shape_item)
labelme_data = {
"version": LABELME_VERSION,
"flags": {},
"shapes": shapes,
"imagePath": os.path.basename(image_path),
"imageData": image_data_b64,
"imageHeight": h,
"imageWidth": w
}
with open(output_json, "w", encoding="utf-8") as f:
json.dump(labelme_data, f, indent=4, ensure_ascii=False)
print(f"转换完成: {output_json}")
def main():
with open(train_txt, "r", encoding="utf-8") as f:
lines = [l.strip() for l in f if l.strip()]
for img_rel_path in lines:
img_path = os.path.join(dataset_path, img_rel_path)
filename = os.path.basename(img_path)
name_only, _ = os.path.splitext(filename)
label_name = name_only + ".txt"
label_path = os.path.join(label_folder, label_name)
output_json_path = os.path.join(output_json_folder, name_only + ".json")
convert_yolo_to_labelme_polygon(img_path, label_path, output_json_path)
if __name__ == "__main__":
main()