Python实现Labelme的Json标注文件与YOLO格式的TXT标注文件相互转换

编程入门行业动态更新时间:2024-10-25 06:22:20

Python实现Labelme的Json标注<a href=https://www.elefans.com/category/jswz/34/1771438.html style= 文件与YOLO格式的TXT标注文件相互转换"/>

Python实现Labelme的Json标注文件与YOLO格式的TXT标注文件相互转换

前言
前提条件
相关介绍
实验环境
Labelme的Json标注文件与YOLO格式的TXT标注文件相互转换
- convert_labelme_json_to_txt
- - jsons/000000000009.json
  - 代码实现
  - 输出结果
  - labels/000000000009.txt
- convert_txt_to_labelme_json
- - labels/000000000009.txt
  - 代码实现
  - 输出结果
  - jsons/000000000009.json
参考

前言

由于本人水平有限，难免出现错漏，敬请批评改正。
更多精彩内容，可点击进入Python日常小操作专栏、OpenCV-Python小应用专栏、YOLO系列专栏、自然语言处理专栏或我的个人主页查看
基于DETR的人脸伪装检测
YOLOv7训练自己的数据集（口罩检测）
YOLOv8训练自己的数据集（足球检测）
YOLOv5：TensorRT加速YOLOv5模型推理
YOLOv5：IoU、GIoU、DIoU、CIoU、EIoU
玩转Jetson Nano（五）：TensorRT加速YOLOv5目标检测
YOLOv5：添加SE、CBAM、CoordAtt、ECA注意力机制
YOLOv5：yolov5s.yaml配置文件解读、增加小目标检测层
Python将COCO格式实例分割数据集转换为YOLO格式实例分割数据集
YOLOv5：使用7.0版本训练自己的实例分割模型（车辆、行人、路标、车道线等实例分割）
使用Kaggle GPU资源免费体验Stable Diffusion开源项目

前提条件

熟悉Python

实验环境

Python 3.x （面向对象的高级语言）

Labelme的Json标注文件与YOLO格式的TXT标注文件相互转换

背景：Labelme仍市面上是一款主流的免费图像标注工具，比如可以用其来进行目标检测、图像分割等的标注，目前，还是有很多数据集标注文件是Labelme格式，因此，实现Labelme的Json标注文件与YOLO格式的TXT标注文件相互转换，是非常有必要的。
目录结构

convert_labelme_json_to_txt

jsons/000000000009.json

{"version": "4.5.7","flags": {},"shapes": [{"label": "45","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[1.0799999999999699,187.69008],[612.66976,473.53008]]},{"label": "45","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[311.73024,4.310160000000001],[631.01024,232.99032]]},{"label": "50","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[249.60032000000007,229.27032],[565.84032,474.35016]]},{"label": "45","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[0.0003200000000092018,13.510080000000002],[434.48032,388.63007999999996]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[376.2,40.35996],[451.75007999999997,86.88996]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[465.7797119999999,38.969952],[523.849728,85.63996800000001]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[385.70016000000004,73.65983999999999],[469.72,144.16992]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[364.04959999999994,2.4900960000000016],[458.80992,73.559856]]}],"imagePath": "000000000009.jpg","imageData": null,"imageHeight": 480,"imageWidth": 640
}

代码实现

import json
import os
import glob
from tqdm import tqdmdef convert_poly_to_rect(coordinateList):X = [int(coordinateList[2 * i]) for i in range(int(len(coordinateList) / 2))]Y = [int(coordinateList[2 * i + 1]) for i in range(int(len(coordinateList) / 2))]Xmax = max(X)Xmin = min(X)Ymax = max(Y)Ymin = min(Y)flag = Falseif (Xmax - Xmin) == 0 or (Ymax - Ymin) == 0:flag = Truereturn [Xmin, Ymin, Xmax - Xmin, Ymax - Ymin], flagdef convert_labelme_json_to_txt(json_path,img_path,out_txt_path):json_list = glob.glob(json_path + '/*.json')num = len(json_list)for json_path in tqdm(json_list):with open(json_path, "r")as f_json:json_data = json.loads(f_json.read())infos = json_data['shapes']if len(infos) ==0:continueimg_w = json_data['imageWidth']img_h = json_data['imageHeight']image_name = json_data['imagePath']image_path = os.path.join(img_path, image_name)if not os.path.exists(img_path):print(img_path, 'is None!')continuetxt_name = os.path.basename(json_path).split('.')[0] + '.txt'txt_path = os.path.join(out_txt_path, txt_name)f = open(txt_path, 'w')for label in infos:points = label['points']if len(points) < 2:continueif len(points) == 2:x1 = points[0][0]y1 = points[0][1]x2 = points[1][0]y2 = points[1][1]points = [[x1, y1], [x2, y1], [x2, y2], [x1, y2]]else:if len(points) < 4:continuesegmentation = []for p in points:segmentation.append(int(p[0]))segmentation.append(int(p[1]))bbox, flag = convert_poly_to_rect(list(segmentation))x1, y1, w, h = bboxif flag:continuex_center = x1 + w/2y_center = y1 + h/2norm_x = x_center / img_wnorm_y = y_center / img_hnorm_w = w / img_wnorm_h = h / img_hobj_cls = label['label']line = [obj_cls, norm_x, norm_y, norm_w, norm_h]line = [str(ll) for ll in line]line = ' '.join(line) + '\n'f.write(line)f.close()if __name__ == "__main__":img_path = 'images'json_path = 'jsons'out_txt_path = 'labels'if not os.path.exists(out_txt_path):os.makedirs(out_txt_path)convert_labelme_json_to_txt(json_path,img_path,out_txt_path)

输出结果

labels/000000000009.txt

45 0.47890625 0.6875 0.9546875 0.5958333333333333
45 0.7359375 0.24583333333333332 0.5 0.475
50 0.6359375 0.7322916666666667 0.49375 0.5104166666666666
45 0.3390625 0.41770833333333335 0.678125 0.78125
49 0.64609375 0.13125 0.1171875 0.09583333333333334
49 0.771875 0.128125 0.090625 0.09791666666666667
49 0.6671875 0.22604166666666667 0.13125 0.14791666666666667
49 0.6421875 0.078125 0.146875 0.14791666666666667

convert_txt_to_labelme_json

labels/000000000009.txt

45 0.47890625 0.6875 0.9546875 0.595833
45 0.7359375 0.24583333333333332 0.5 0.475
50 0.6359375 0.7322916666666667 0.49375 0.5104166
45 0.3390625 0.41770833333333335 0.678125 0.78125
49 0.64609375 0.13125 0.1171875 0.095833
49 0.771875 0.128125 0.090625 0.0979166
49 0.6671875 0.22604166666666667 0.13125 0.1479166
49 0.6421875 0.078125 0.146875 0.1479166

代码实现

import os
import cv2
import json
import glob
import numpy as npdef convert_txt_to_labelme_json(txt_path, image_path, output_dir, image_fmt='.jpg'):# txt 转labelme json# 将yolo的txt转labelme jsontxts = glob.glob(os.path.join(txt_path, "*.txt"))for txt in txts:labelme_json = {'version': '4.5.7','flags': {},'shapes': [],'imagePath': None,'imageData': None,'imageHeight': None,'imageWidth': None,}txt_name = os.path.basename(txt)image_name = txt_name.split(".")[0] + image_fmtlabelme_json['imagePath'] = image_nameimage_name = os.path.join(image_path, image_name)if not os.path.exists(image_name):raise Exception('txt 文件={},找不到对应的图像={}'.format(txt, image_name))image = cv2.imdecode(np.fromfile(image_name, dtype=np.uint8), cv2.IMREAD_COLOR)h, w = image.shape[:2]labelme_json['imageHeight'] = hlabelme_json['imageWidth'] = wwith open(txt, 'r') as t:lines = t.readlines()for line in lines:content = line.split(' ')label = content[0]object_width = float(content[3])object_height = float(content[4])top_left_x = (float(content[1]) - object_width / 2) * wtop_left_y = (float(content[2]) - object_height / 2) * hbottom_right_x = (float(content[1]) + object_width / 2) * wbottom_right_y = (float(content[2]) + object_height / 2) * htry:shape = {'label': str(label),'score':float(content[5]),'group_id': None,'shape_type': 'rectangle','flags': {},'points': [[float(top_left_x), float(top_left_y)],[float(bottom_right_x), float(bottom_right_y)]]}except Exception as e:# print(e)shape = {'label': str(label),'score':float(0.99),'group_id': None,'shape_type': 'rectangle','flags': {},'points': [[float(top_left_x), float(top_left_y)],[float(bottom_right_x), float(bottom_right_y)]]}labelme_json['shapes'].append(shape)json_name = txt_name.split('.')[0] + '.json'json_name_path = os.path.join(output_dir, json_name)fd = open(json_name_path, 'w')json.dump(labelme_json, fd, indent=4)fd.close()print("save json={}".format(json_name_path))if __name__=="__main__":in_imgs_dir = 'images'in_label_txt_dir = 'labels'out_labelme_json_dir = 'jsons'if not os.path.exists(out_labelme_json_dir):os.mkdir(out_labelme_json_dir)convert_txt_to_labelme_json(in_label_txt_dir,in_imgs_dir,out_labelme_json_dir,image_fmt='.jpg')

输出结果

jsons/000000000009.json

{"version": "4.5.7","flags": {},"shapes": [{"label": "45","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[1.0799999999999699,187.69008],[612.66976,473.53008]]},{"label": "45","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[311.73024,4.310160000000001],[631.01024,232.99032]]},{"label": "50","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[249.60032000000007,229.27032],[565.84032,474.35016]]},{"label": "45","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[0.0003200000000092018,13.510080000000002],[434.48032,388.63007999999996]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[376.2,40.35996],[451.75007999999997,86.88996]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[465.7797119999999,38.969952],[523.849728,85.63996800000001]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[385.70016000000004,73.65983999999999],[469.72,144.16992]]},{"label": "49","score": 0.99,"group_id": null,"shape_type": "rectangle","flags": {},"points": [[364.04959999999994,2.4900960000000016],[458.80992,73.559856]]}],"imagePath": "000000000009.jpg","imageData": null,"imageHeight": 480,"imageWidth": 640
}

参考

[1]
[2]

由于本人水平有限，难免出现错漏，敬请批评改正。
更多精彩内容，可点击进入Python日常小操作专栏、OpenCV-Python小应用专栏、YOLO系列专栏、自然语言处理专栏或我的个人主页查看
基于DETR的人脸伪装检测
YOLOv7训练自己的数据集（口罩检测）
YOLOv8训练自己的数据集（足球检测）
YOLOv5：TensorRT加速YOLOv5模型推理
YOLOv5：IoU、GIoU、DIoU、CIoU、EIoU
玩转Jetson Nano（五）：TensorRT加速YOLOv5目标检测
YOLOv5：添加SE、CBAM、CoordAtt、ECA注意力机制
YOLOv5：yolov5s.yaml配置文件解读、增加小目标检测层
Python将COCO格式实例分割数据集转换为YOLO格式实例分割数据集
YOLOv5：使用7.0版本训练自己的实例分割模型（车辆、行人、路标、车道线等实例分割）
使用Kaggle GPU资源免费体验Stable Diffusion开源项目