python3GUI--基于YOLO的火焰与烟雾检测系统By:PyQt5（详细图文介绍）

import os
import shutil
import random

# 输入路径
labels_dir = r'C:\Users\13978\Downloads\fire_detection\VOC2020\labels'           # 输入YOLO标签文件目录
images_dir = r'C:\Users\13978\Downloads\fire_detection\VOC2020\JPEGImages'           # 对应图像文件目录
output_dir = r'C:\Users\13978\Downloads\fire_detection\VOC2020\output_dataset2'   # 输出目录

# 分配数量
train_count = 400
val_count = 100

# 创建输出目录结构
for split in ['train', 'val']:
    os.makedirs(os.path.join(output_dir, split, 'images'), exist_ok=True)
    os.makedirs(os.path.join(output_dir, split, 'labels'), exist_ok=True)

# 获取所有标注文件（假设图像和标签同名）
label_files = [f for f in os.listdir(labels_dir) if f.endswith('.txt')]
random.shuffle(label_files)

# 检查数量是否足够
total_needed = train_count + val_count
if len(label_files) < total_needed:
    raise ValueError(f"标签文件不足，共有 {len(label_files)} 个，要求至少 {total_needed} 个。")

# 分配训练和验证
train_files = label_files[:train_count]
val_files = label_files[train_count:train_count + val_count]

def copy_samples(file_list, split_name):
    for label_file in file_list:
        base_name = os.path.splitext(label_file)[0]
        image_file = None

        # 尝试匹配图像扩展名
        for ext in ['.jpg', '.png', '.jpeg']:
            possible_image = os.path.join(images_dir, base_name + ext)
            if os.path.exists(possible_image):
                image_file = possible_image
                break

        if image_file is None:
            print(f"⚠️ 图像文件 {base_name} 不存在，跳过。")
            continue

        # 拷贝标签和图像
        shutil.copy(os.path.join(labels_dir, label_file), os.path.join(output_dir, split_name, 'labels', label_file))
        shutil.copy(image_file, os.path.join(output_dir, split_name, 'images', os.path.basename(image_file)))

    print(f"✅ {split_name} 集合完成，共复制 {len(file_list)} 张图像。")

# 执行复制
copy_samples(train_files, 'train')
copy_samples(val_files, 'val')

import os
import xml.etree.ElementTree as ET
from PIL import Image

# VOC 类别列表（顺序必须固定）
classes = ['fire']  # 替换成你自己的类别

# 输入路径
voc_anno_dir = r'C:\Users\13978\Downloads\fire_detection\VOC2020\Annotations'
voc_img_dir = r'C:\Users\13978\Downloads\fire_detection\VOC2020\JPEGImages'

# 输出路径
yolo_label_dir = r'C:\Users\13978\Downloads\fire_detection\VOC2020\labels'
os.makedirs(yolo_label_dir, exist_ok=True)


def convert_box(size, box):
    """将VOC边界框坐标转为YOLO格式"""
    dw = 1. / size[0]
    dh = 1. / size[1]
    x_center = (box[0] + box[1]) / 2.0
    y_center = (box[2] + box[3]) / 2.0
    w = box[1] - box[0]
    h = box[3] - box[2]
    return (x_center * dw, y_center * dh, w * dw, h * dh)


def convert_annotation(xml_path, img_path, output_path):
    tree = ET.parse(xml_path)
    root = tree.getroot()
    img = Image.open(img_path)
    w, h = img.size

    with open(output_path, 'w') as out_file:
        for obj in root.iter('object'):
            cls = obj.find('name').text
            if cls not in classes:
                continue
            cls_id = classes.index(cls)
            xml_box = obj.find('bndbox')
            b = (
                float(xml_box.find('xmin').text),
                float(xml_box.find('xmax').text),
                float(xml_box.find('ymin').text),
                float(xml_box.find('ymax').text)
            )
            bb = convert_box((w, h), b)
            out_file.write(f"{cls_id} {' '.join(map(str, bb))}\n")


# 遍历VOC标注文件夹
for filename in os.listdir(voc_anno_dir):
    if not filename.endswith('.xml'):
        continue
    xml_path = os.path.join(voc_anno_dir, filename)
    img_filename = filename.replace('.xml', '.jpg')
    img_path = os.path.join(voc_img_dir, img_filename)
    output_path = os.path.join(yolo_label_dir, filename.replace('.xml', '.txt'))

    if not os.path.exists(img_path):
        print(f"Warning: image {img_path} not found, skipping.")
        continue

    convert_annotation(xml_path, img_path, output_path)
    print(f"Converted {filename} to {output_path}")

path: C:\Users\13978\Downloads\fire_detection\VOC2020\output_dataset2  # 数据集的根目录
train: train                 # 相对于 path 的训练图像路径
val: val                     # 相对于 path 的验证图像路径

names:
  0: smoke
  1: fire

yolo task=detect mode=train model=../data/models/yolov8n.pt data=./data.yaml epochs=50 imgsz=640  batch=16 lr0=0.01

卷 Data 的文件夹 PATH 列表
卷序列号为 9AD9-E265
D:\PROJECTS\GITEE\YOLO_PROJECTS\YOLOV8-FIRE-SMOKE-DETECTION\TRAIN\RUNS\DETECT\TRAIN3
│  args.yaml
│  confusion_matrix.png
│  confusion_matrix_normalized.png
│  events.out.tfevents.1753497631.Admin.16932.0
│  F1_curve.png
│  labels.jpg
│  labels_correlogram.jpg
│  PR_curve.png
│  P_curve.png
│  results.csv
│  results.png
│  R_curve.png
│  train_batch0.jpg
│  train_batch1.jpg
│  train_batch1720.jpg
│  train_batch1721.jpg
│  train_batch1722.jpg
│  train_batch2.jpg
│  val_batch0_labels.jpg
│  val_batch0_pred.jpg
│  val_batch1_labels.jpg
│  val_batch1_pred.jpg
│  val_batch2_labels.jpg
│  val_batch2_pred.jpg
│
└─weights
        best.pt
        last.pt

PyQt5==5.15.11
PyQt5_sip==12.15.0
QtAwesome==1.3.1
torch==2.4.1
torchvision==0.19.1
Pillow==9.3.0
PyQtWebEngine==5.15.5
opencv-python==4.10.0.82
ultralytics==8.3.98
pyttsx3==2.71

pip install -r requirements.txt