openmmlab-mmdet

json2coco

import json

import re

from collections import defaultdict

import cv2

import os

# 读取JSON文件

with open(r'F:\openmmlab\mmdetection-main\data\balloon_dataset\balloon\val\via_region_data.json') as f:

    annotations = json.load(f)

# 初始化COCO格式字典

coco_dict = defaultdict(list)

coco_dict['info'] = {}

coco_dict['licenses'] = []

coco_dict['categories'] = []

coco_dict['images'] = []

coco_dict['annotations'] = []

# 添加info信息

coco_dict['info']['version'] = '1.0'

coco_dict['info']['description'] = 'Balloon Dataset'

coco_dict['info']['year'] = 2023

coco_dict['info']['contributor'] = 'YUJUNYU'

coco_dict['info']['date_created'] = '2023-06-09'

# 添加categories信息

category_id = 1

category = {'id': category_id, 'name': 'balloon', 'supercategory': 'object'}

coco_dict['categories'].append(category)

# 遍历每个图像的注释信息

annotation_id = 1

for image_id, image_info in annotations.items():

    # 从image_id中提取出正确的图像ID

    match = re.search(r'(\d+)_.*', image_id)

    if match:

        image_id = match.group(1)

    else:

        continue

   

    # 获取图像的文件名和大小

    filename = image_info['filename']

    size = image_info['size']

    # 获取图像的height和width

    img = cv2.imread(os.path.join(r'F:\openmmlab\mmdetection-main\data\balloon_dataset\balloon\val', filename))

    height, width, _ = img.shape

    # print(filename,height,width)

   

    # 添加图像信息

    image = {'id': int(image_id), 'file_name': filename, 'width': width, 'height': height}

    coco_dict['images'].append(image)

   

    # 获取多边形注释信息

    regions = image_info['regions']

    for region_id, region_info in regions.items():

        # 获取多边形顶点坐标

        x = region_info['shape_attributes']['all_points_x']

        y = region_info['shape_attributes']['all_points_y']

        points = list(zip(x, y))

       

        # 计算多边形的边界框

        xmin = min(x)

        xmax = max(x)

        ymin = min(y)

        ymax = max(y)

        width = xmax - xmin

        height = ymax - ymin

       

        # 添加注释信息

        annotation = {'id': annotation_id, 'image_id': int(image_id), 'category_id': category_id, 'segmentation': [points], 'area': width * height, 'bbox': [xmin, ymin, width, height], 'iscrowd': 0}

        coco_dict['annotations'].append(annotation)

        annotation_id += 1

# 保存COCO格式文件

with open(r'F:\openmmlab\mmdetection-main\data\balloon_dataset\balloon\val\balloon_val.json', 'w') as f:

    json.dump(coco_dict, f)

print('success！')