在自己的数据集上实验时，往往需要将VOC数据集转化为coco数据集，因为这种需求所以才记录这篇文章，代码出处未知，感谢开源。
在远程服务器上测试目标检测算法需要用到测试集，最常用的是coco2014/2017和voc07/12数据集。
coco数据集的地址为http://cocodataset.org/#download
voc和coco的镜像为https://pjreddie.com/projects/pascal-voc-dataset-mirror/
一、数据集格式对比
1.1 VOC数据集
VOC_ROOT #根目录 ├── JPEGImages # 存放源图，(当然图片并不一定要是**.jpg格式的，只是规定文件夹名字叫JPEGImages**)； │ ├── aaaa.jpg │ ├── bbbb.jpg │ └── cccc.jpg ├── Annotations # 存放xml文件，VOC的标注是xml格式，与JPEGImages中的图片一一对应 │ ├── aaaa.xml │ ├── bbbb.xml │ └── cccc.xml └── ImageSets └── Main ├── train.txt # txt文件中每一行包含一个图片的名称 └── val.txt
1.2 COCO数据集
COCO_ROOT #根目录 ├── annotations # 存放json格式的标注 │ ├── instances_train2017.json │ └── instances_val2017.json └── train2017 # 存放图片文件 │ ├── 000000000001.jpg │ ├── 000000000002.jpg │ └── 000000000003.jpg └── val2017 ├── 000000000004.jpg └── 000000000005.jpg
1.2.3 json标注格式
与VOC一个文件一个xml标注不同，COCO所有的目标框标注都是放在一个json文件中的。
这个json文件解析出来是一个字典，格式如下：
{ "info": info, "images": [image], "annotations": [annotation], "categories": [categories], "licenses": [license], }
二、转换步骤
2.1 程序总体目录
2.2 标签文件转换代码实现（xml文件转json格式）VOC_To_CoCo_01.py
这里需要运行三次，因为train.txt val.txt test.txt是三个文件，具体看注释
# VOC_To_CoCo_01.py import os import argparse import json import xml.etree.ElementTree as ET from typing import Dict, List import re def get_label2id(labels_path: str) -> Dict[str, int]: """id is 1 start""" with open(labels_path, 'r') as f: labels_str = f.read().split() labels_ids = list(range(1, len(labels_str) + 1)) return dict(zip(labels_str, labels_ids)) def get_annpaths(ann_dir_path: str = None, ann_ids_path: str = None, ext: str = '', annpaths_list_path: str = None) -> List[str]: # If use annotation paths list if annpaths_list_path is not None: with open(annpaths_list_path, 'r') as f: ann_paths = f.read().split() return ann_paths # If use annotaion ids list ext_with_dot = '.' + ext if ext != '' else '' with open(ann_ids_path, 'r') as f: ann_ids = f.read().split() ann_paths = [os.path.join(ann_dir_path, aid + ext_with_dot) for aid in ann_ids] return ann_paths def get_image_info(annotation_root, extract_num_from_imgid=True): path = annotation_root.findtext('path') if path is None: filename = annotation_root.findtext('filename') else: filename = os.path.basename(path) img_name = os.path.basename(filename) img_id = os.path.splitext(img_name)[0] if extract_num_from_imgid and isinstance(img_id, str): img_id = int(re.findall(r'\d+', img_id)[0]) size = annotation_root.find('size') width = int(size.findtext('width')) height = int(size.findtext('height')) image_info = { 'file_name': filename, 'height': height, 'width': width, 'id': img_id } return image_info def get_coco_annotation_from_obj(obj, label2id): label = obj.findtext('name') assert label in label2id, f"Error: {label} is not in label2id !" category_id = label2id[label] bndbox = obj.find('bndbox') xmin = int(bndbox.findtext('xmin')) - 1 ymin = int(bndbox.findtext('ymin')) - 1 xmax = int(bndbox.findtext('xmax')) ymax = int(bndbox.findtext('ymax')) assert xmax > xmin and ymax > ymin, f"Box size error !: (xmin, ymin, xmax, ymax): {xmin, ymin, xmax, ymax}" o_width = xmax - xmin o_height = ymax - ymin ann = { 'area': o_width * o_height, 'iscrowd': 0, 'bbox': [xmin, ymin, o_width, o_height], 'category_id': category_id, 'ignore': 0, 'segmentation': [] # This script is not for segmentation } return ann def convert_xmls_to_cocojson(annotation_paths: List[str], label2id: Dict[str, int], output_jsonpath: str, extract_num_from_imgid: bool = True): output_json_dict = { "images": [], "type": "instances", "annotations": [], "categories": [] } bnd_id = 1 # START_BOUNDING_BOX_ID, TODO input as args ? for a_path in annotation_paths: # Read annotation xml ann_tree = ET.parse(a_path) ann_root = ann_tree.getroot() img_info = get_image_info(annotation_root=ann_root, extract_num_from_imgid=extract_num_from_imgid) img_id = img_info['id'] output_json_dict['images'].append(img_info) for obj in ann_root.findall('object'): ann = get_coco_annotation_from_obj(obj=obj, label2id=label2id) ann.update({'image_id': img_id, 'id': bnd_id}) output_json_dict['annotations'].append(ann) bnd_id = bnd_id + 1 for label, label_id in label2id.items(): category_info = {'supercategory': 'none', 'id': label_id, 'name': label} output_json_dict['categories'].append(category_info) with open(output_jsonpath, 'w') as f: output_json = json.dumps(output_json_dict) f.write(output_json) print('Convert successfully !') def main(): parser = argparse.ArgumentParser( description='This script support converting voc format xmls to coco format json') parser.add_argument('--ann_dir', type=str, default='./VOCdevkit/Annotations') parser.add_argument('--ann_ids', type=str, default='./VOCdevkit/ImageSets/Main/val.txt') # 这里修改 train val test 一共修改三次 #parser.add_argument('--ann_ids', type=str, default='./VOCdevkit/ImageSets/Main/train.txt') #parser.add_argument('--ann_ids', type=str, default='./VOCdevkit/ImageSets/Main/test.txt') parser.add_argument('--ann_paths_list', type=str, default=None) parser.add_argument('--labels', type=str, default='./VOCdevkit/labels.txt') parser.add_argument('--output', type=str, default='./output/annotations/val.json') # 这里修改 train val test 一共修改三次 #parser.add_argument('--output', type=str, default='./output/annotations/train.json') #parser.add_argument('--output', type=str, default='./output/annotations/test.json') parser.add_argument('--ext', type=str, default='xml') args = parser.parse_args() label2id = get_label2id(labels_path=args.labels) ann_paths = get_annpaths( ann_dir_path=args.ann_dir, ann_ids_path=args.ann_ids, ext=args.ext, annpaths_list_path=args.ann_paths_list ) convert_xmls_to_cocojson( annotation_paths=ann_paths, label2id=label2id, output_jsonpath=args.output, extract_num_from_imgid=True ) if __name__ == '__main__': if not os.path.exists('./output/annotations'): os.makedirs('./output/annotations') main()
2.3 数据集图像文件copy代码实现（复制图片数据集到coco中）VOC_To_CoCo_02.py
# VOC_To_CoCo_02.py import os import shutil images_file_path = './VOCdevkit/JPEGImages/' split_data_file_path = './VOCdevkit/ImageSets/Main/' new_images_file_path = './output/' if not os.path.exists(new_images_file_path + 'train'): os.makedirs(new_images_file_path + 'train') if not os.path.exists(new_images_file_path + 'val'): os.makedirs(new_images_file_path + 'val') if not os.path.exists(new_images_file_path + 'test'): os.makedirs(new_images_file_path + 'test') dst_train_Image = new_images_file_path + 'train/' dst_val_Image = new_images_file_path + 'val/' dst_test_Image = new_images_file_path + 'test/' total_txt = os.listdir(split_data_file_path) for i in total_txt: name = i[:-4] if name == 'train': txt_file = open(split_data_file_path + i, 'r') for line in txt_file: line = line.strip('\n') line = line.strip('\r') srcImage = images_file_path + line + '.jpg' dstImage = dst_train_Image + line + '.jpg' shutil.copyfile(srcImage, dstImage) txt_file.close() elif name == 'val': txt_file = open(split_data_file_path + i, 'r') for line in txt_file: line = line.strip('\n') line = line.strip('\r') srcImage = images_file_path + line + '.jpg' dstImage = dst_val_Image + line + '.jpg' shutil.copyfile(srcImage, dstImage) txt_file.close() elif name == 'test': txt_file = open(split_data_file_path + i, 'r') for line in txt_file: line = line.strip('\n') line = line.strip('\r') srcImage = images_file_path + line + '.jpg' dstImage = dst_test_Image + line + '.jpg' shutil.copyfile(srcImage, dstImage) txt_file.close() else: print("Error, Please check the file name of folder")
三、效果展示
