当前位置： 首页 > news >正文

split_conversion将json转成yolo训练用的txt,在直接按照8：1：1的比例分成训练集，测试集，验证集

news 来源：原创 2025/6/1 11:36:38

第一章  使用说明

类别自己在代码中改，其他四个参数

--json-folder:json文件夹路径

--txt-folder：转换成功后txt的存放路径

--images-dir：图片文件夹路径

--save-dir：转换完成分割后所有文件的路径

终端命令行：python split_conversion.py --json-folder "" --txt-folder "" --images-dir "" --save-dir ""

""处是四个路径

第二章 导包

import json
import os
from tqdm import tqdm
import shutil
import random
import argparse

第三章 转换格式函数

def labelme_json_to_yolo(txt_save_path, json_path, class_map):"""将LabelMe格式JSON转换为YOLO TXT标注文件:param txt_save_path: TXT保存路径:param json_path: LabelMe JSON文件路径:param class_map: 类别名映射字典（如{"person": 0}）"""with open(json_path, 'r', encoding='utf-8') as f:data = json.load(f)img_height = data["imageHeight"]img_width = data["imageWidth"]img_name = os.path.basename(data["imagePath"])  # 提取图片文件名txt_name = os.path.splitext(img_name)[0] + ".txt"txt_path = os.path.join(txt_save_path, txt_name)lines = []for shape in data["shapes"]:label = shape["label"]if label not in class_map:continue  # 跳过未定义类别cls_id = class_map[label]# 解析多边形/矩形标注为边界框（xmin, ymin, xmax, ymax）points = shape["points"]x_coords = [p[0] for p in points]y_coords = [p[1] for p in points]xmin = min(x_coords)ymin = min(y_coords)xmax = max(x_coords)ymax = max(y_coords)# 计算YOLO格式坐标（归一化中心坐标+宽高）x_center = (xmin + xmax) / (2 * img_width)y_center = (ymin + ymax) / (2 * img_height)w = (xmax - xmin) / img_widthh = (ymax - ymin) / img_height# 过滤无效坐标（避免越界）if 0 <= x_center <= 1 and 0 <= y_center <= 1 and w > 0 and h > 0:lines.append(f"{cls_id} {x_center:.6f} {y_center:.6f} {w:.6f} {h:.6f}")# 保存TXT文件with open(txt_path, 'w') as f:f.write('\n'.join(lines))

第四章 划分函数

def batch_convert_labelme_to_yolo(json_folder, txt_folder, class_map):"""批量转换LabelMe格式JSON文件夹到YOLO TXT:param json_folder: JSON文件夹路径:param txt_folder: TXT保存路径:param class_map: 类别名映射字典（如{"person": 0}）"""os.makedirs(txt_folder, exist_ok=True)json_files = [f for f in os.listdir(json_folder) if f.lower().endswith('.json')]for json_file in tqdm(json_files, desc="转换中"):json_path = os.path.join(json_folder, json_file)labelme_json_to_yolo(txt_folder, json_path, class_map)print(f"批量转换完成！共处理{len(json_files)}个JSON文件，保存到：{txt_folder}")

第五章 确认文件夹是否存在

def mkdir(path):if not os.path.exists(path):os.makedirs(path)

第六章主函数

def main(image_dir, txt_dir, save_dir):# 创建文件夹mkdir(save_dir)images_dir = os.path.join(save_dir, 'images')labels_dir = os.path.join(save_dir, 'labels')img_train_path = os.path.join(images_dir, 'train')img_test_path = os.path.join(images_dir, 'test')img_val_path = os.path.join(images_dir, 'val')label_train_path = os.path.join(labels_dir, 'train')label_test_path = os.path.join(labels_dir, 'test')label_val_path = os.path.join(labels_dir, 'val')mkdir(images_dir)mkdir(labels_dir)mkdir(img_train_path)mkdir(img_test_path)mkdir(img_val_path)mkdir(label_train_path)mkdir(label_test_path)mkdir(label_val_path)# 数据集划分比例，训练集80%，验证集10%，测试集10%，按需修改train_percent = 0.8val_percent = 0.1test_percent = 0.1total_txt = os.listdir(txt_dir)num_txt = len(total_txt)list_all_txt = range(num_txt)  # 范围 range(0, num)num_train = int(num_txt * train_percent)num_val = int(num_txt * val_percent)num_test = num_txt - num_train - num_valtrain = random.sample(list_all_txt, num_train)# 在全部数据集中取出trainval_test = [i for i in list_all_txt if not i in train]# 再从val_test取出num_val个元素，val_test剩下的元素就是testval = random.sample(val_test, num_val)print("训练集数目：{}, 验证集数目：{}, 测试集数目：{}".format(len(train), len(val), len(val_test) - len(val)))for i in list_all_txt:name = total_txt[i][:-4]srcImage = os.path.join(image_dir, name + '.jpg')srcLabel = os.path.join(txt_dir, name + '.txt')if i in train:dst_train_Image = os.path.join(img_train_path, name + '.jpg')dst_train_Label = os.path.join(label_train_path, name + '.txt')shutil.copyfile(srcImage, dst_train_Image)shutil.copyfile(srcLabel, dst_train_Label)elif i in val:dst_val_Image = os.path.join(img_val_path, name + '.jpg')dst_val_Label = os.path.join(label_val_path, name + '.txt')shutil.copyfile(srcImage, dst_val_Image)shutil.copyfile(srcLabel, dst_val_Label)else:dst_test_Image = os.path.join(img_test_path, name + '.jpg')dst_test_Label = os.path.join(label_test_path, name + '.txt')shutil.copyfile(srcImage, dst_test_Image)shutil.copyfile(srcLabel, dst_test_Label)

 第七章 主函数调用

if __name__ == "__main__":parser = argparse.ArgumentParser(description='Convert LabelMe JSON to YOLO TXT and split datasets')parser.add_argument('--json-folder', type=str, default=r'',help='LabelMe JSON folder path')parser.add_argument('--txt-folder', type=str, default=r'',help='YOLO TXT save path')parser.add_argument('--class-map', default={"自己的类别": 0}, type=dict,help='Class name mapping dictionary (e.g. {"person": 0})')parser.add_argument('--images-dir', type=str, default=r'', help='images path dir')parser.add_argument('--save-dir', default=r'', type=str, help='save dir')args = parser.parse_args()json_folder = args.json_foldertxt_folder = args.txt_folderclass_map = args.class_mapimage_dir = args.images_dirsave_dir = args.save_dir# 转换LabelMe JSON到YOLO TXTbatch_convert_labelme_to_yolo(json_folder, txt_folder, class_map)# 划分数据集main(image_dir, txt_folder, save_dir)

全部代码如下： 

import json
import os
from tqdm import tqdm  # 可选进度条库
import shutil
import random
import argparsedef labelme_json_to_yolo(txt_save_path, json_path, class_map):"""将LabelMe格式JSON转换为YOLO TXT标注文件:param txt_save_path: TXT保存路径:param json_path: LabelMe JSON文件路径:param class_map: 类别名映射字典（如{"person": 0}）"""with open(json_path, 'r', encoding='utf-8') as f:data = json.load(f)img_height = data["imageHeight"]img_width = data["imageWidth"]img_name = os.path.basename(data["imagePath"])  # 提取图片文件名txt_name = os.path.splitext(img_name)[0] + ".txt"txt_path = os.path.join(txt_save_path, txt_name)lines = []for shape in data["shapes"]:label = shape["label"]if label not in class_map:continue  # 跳过未定义类别cls_id = class_map[label]# 解析多边形/矩形标注为边界框（xmin, ymin, xmax, ymax）points = shape["points"]x_coords = [p[0] for p in points]y_coords = [p[1] for p in points]xmin = min(x_coords)ymin = min(y_coords)xmax = max(x_coords)ymax = max(y_coords)# 计算YOLO格式坐标（归一化中心坐标+宽高）x_center = (xmin + xmax) / (2 * img_width)y_center = (ymin + ymax) / (2 * img_height)w = (xmax - xmin) / img_widthh = (ymax - ymin) / img_height# 过滤无效坐标（避免越界）if 0 <= x_center <= 1 and 0 <= y_center <= 1 and w > 0 and h > 0:lines.append(f"{cls_id} {x_center:.6f} {y_center:.6f} {w:.6f} {h:.6f}")# 保存TXT文件with open(txt_path, 'w') as f:f.write('\n'.join(lines))def batch_convert_labelme_to_yolo(json_folder, txt_folder, class_map):"""批量转换LabelMe格式JSON文件夹到YOLO TXT:param json_folder: JSON文件夹路径:param txt_folder: TXT保存路径:param class_map: 类别名映射字典（如{"person": 0}）"""os.makedirs(txt_folder, exist_ok=True)json_files = [f for f in os.listdir(json_folder) if f.lower().endswith('.json')]for json_file in tqdm(json_files, desc="转换中"):json_path = os.path.join(json_folder, json_file)labelme_json_to_yolo(txt_folder, json_path, class_map)print(f"批量转换完成！共处理{len(json_files)}个JSON文件，保存到：{txt_folder}")# 检查文件夹是否存在
def mkdir(path):if not os.path.exists(path):os.makedirs(path)def main(image_dir, txt_dir, save_dir):# 创建文件夹mkdir(save_dir)images_dir = os.path.join(save_dir, 'images')labels_dir = os.path.join(save_dir, 'labels')img_train_path = os.path.join(images_dir, 'train')img_test_path = os.path.join(images_dir, 'test')img_val_path = os.path.join(images_dir, 'val')label_train_path = os.path.join(labels_dir, 'train')label_test_path = os.path.join(labels_dir, 'test')label_val_path = os.path.join(labels_dir, 'val')mkdir(images_dir)mkdir(labels_dir)mkdir(img_train_path)mkdir(img_test_path)mkdir(img_val_path)mkdir(label_train_path)mkdir(label_test_path)mkdir(label_val_path)# 数据集划分比例，训练集80%，验证集10%，测试集10%，按需修改train_percent = 0.8val_percent = 0.1test_percent = 0.1total_txt = os.listdir(txt_dir)num_txt = len(total_txt)list_all_txt = range(num_txt)  # 范围 range(0, num)num_train = int(num_txt * train_percent)num_val = int(num_txt * val_percent)num_test = num_txt - num_train - num_valtrain = random.sample(list_all_txt, num_train)# 在全部数据集中取出trainval_test = [i for i in list_all_txt if not i in train]# 再从val_test取出num_val个元素，val_test剩下的元素就是testval = random.sample(val_test, num_val)print("训练集数目：{}, 验证集数目：{}, 测试集数目：{}".format(len(train), len(val), len(val_test) - len(val)))for i in list_all_txt:name = total_txt[i][:-4]srcImage = os.path.join(image_dir, name + '.jpg')srcLabel = os.path.join(txt_dir, name + '.txt')if i in train:dst_train_Image = os.path.join(img_train_path, name + '.jpg')dst_train_Label = os.path.join(label_train_path, name + '.txt')shutil.copyfile(srcImage, dst_train_Image)shutil.copyfile(srcLabel, dst_train_Label)elif i in val:dst_val_Image = os.path.join(img_val_path, name + '.jpg')dst_val_Label = os.path.join(label_val_path, name + '.txt')shutil.copyfile(srcImage, dst_val_Image)shutil.copyfile(srcLabel, dst_val_Label)else:dst_test_Image = os.path.join(img_test_path, name + '.jpg')dst_test_Label = os.path.join(label_test_path, name + '.txt')shutil.copyfile(srcImage, dst_test_Image)shutil.copyfile(srcLabel, dst_test_Label)if __name__ == "__main__":parser = argparse.ArgumentParser(description='Convert LabelMe JSON to YOLO TXT and split datasets')parser.add_argument('--json-folder', type=str, default=r'',help='LabelMe JSON folder path')parser.add_argument('--txt-folder', type=str, default=r'',help='YOLO TXT save path')parser.add_argument('--class-map', default={"自己的类别": 0}, type=dict,help='Class name mapping dictionary (e.g. {"person": 0})')parser.add_argument('--images-dir', type=str, default=r'', help='images path dir')parser.add_argument('--save-dir', default=r'', type=str, help='save dir')args = parser.parse_args()json_folder = args.json_foldertxt_folder = args.txt_folderclass_map = args.class_mapimage_dir = args.images_dirsave_dir = args.save_dir# 转换LabelMe JSON到YOLO TXTbatch_convert_labelme_to_yolo(json_folder, txt_folder, class_map)# 划分数据集main(image_dir, txt_folder, save_dir)

相关文章：

  • 实现一个免费可用的文生图的MCP Server
  • Spring Boot 中 @RequestParam 和 @RequestPart 的区别详解（含实际项目案例）
  • 长短期记忆（LSTM）网络模型
  • 408考研逐题详解：2009年第25题
  • GSR 手环能耗数据实测：STM32 与 SD NAND 的功耗优化成果
  • 智橙PLM与MES系统集成项目执行记录 智渤慧晟机械装备技术服务部 24.08
  • 华锐视点助力，虚拟旅游绽放更璀璨光彩​
  • 视频压制(Video Encoding/Compression）
  • 华为交换机S12708常用命令
  • html+css+js趣味小游戏~Treasure Arena多人竞技（附源码）
  • 杨传辉：构建 Data × AI 能力，打造 AI 时代的一体化数据底座｜OceanBase 开发者大会实录
  • day024-网络基础-TCP与UDP、DNS
  • selenium基础
  • cc攻击是什么？云上业务如何防护cc攻击
  • 【FlashRAG】本地部署与demo运行（二）
  • If possible, you should set the Secure flag for these cookies 修复方案
  • leetcode1201. 丑数 III -medium
  • weibo_comment_pc_tool: 微博评论采集软件工具，根据wb帖子链接爬评论
  • 调用Gensim库训练Word2Vec模型
  • Predixy的docker化
  • wordpress 抄袭查询/朝阳网站seo
  • 武汉成交型网站/seo建站的步骤
  • 成都专业网站建设费用/百度关键词排名工具
  • 新手做网站流程/百度官网app
  • 做网站什么什么/十大收益最好的自媒体平台
  • 进入百度app查看/seo网站优化优化排名