多个xml文件转coco格式、coco转VOC格式（xml）

1、多个xml文件转coco格式

bug已改，已经试过可以用。

只需要修改三个地方：

1. 将 xml_folder 变量中的路径改为你自己的 xml 文件所在文件夹路径。

2. 将 class_name 变量中的类别名称改为你自己数据集的类别名称。

3. 将 json_file 变量中的路径改为你想要保存 COCO 格式注释文件的路径。

然后按住shift+鼠标右键，选择在此处打开powershell，输入 python xxx.py即可转换成功

import logging
import os
import time
import xml.etree.ElementTree as ET
import json

def get_file_list(path, type=".xml"):
    file_names = []
    for maindir, subdir, file_name_list in os.walk(path):
        for filename in file_name_list:
            apath = os.path.join(maindir, filename)
            ext = os.path.splitext(apath)[1]
            if ext == type:
                file_names.append(filename)
    return file_names


def xml_to_coco(ann_path, class_names):
    """
    convert xml annotations to coco_api
    :param ann_path:
    :return:
    """
    class_name_dict = dict(zip(class_names, range(1, len(class_names)+1)))
    print("loading annotations into memory...")
    tic = time.time()
    ann_file_names = get_file_list(ann_path, type=".xml")
    logging.info("Found {} annotation files.".format(len(ann_file_names)))
    image_info = []
    categories = []
    annotations = []
    for idx, supercat in enumerate(class_names):
        categories.append(
            {"supercategory": supercat, "id": idx + 1, "name": supercat}
        )
    ann_id = 1
    for idx, xml_name in enumerate(ann_file_names):
        tree = ET.parse(os.path.join(ann_path, xml_name))
        root = tree.getroot()
        file_name = root.find("filename").text
        width = int(root.find("size").find("width").text)
        height = int(root.find("size").find("height").text)
        info = {
            "file_name": file_name,
            "height": height,
            "width": width,
            "id": idx + 1,
        }
        image_info.append(info)
        for _object in root.findall("object"):
            category = _object.find("name").text
            if category not in class_names:
                print(
                    "WARNING! {} is not in class_names! "
                    "Pass this box annotation.".format(category)
                )
                continue

            cat_id = class_name_dict[category]

            xmin = int(_object.find("bndbox").find("xmin").text)
            ymin = int(_object.find("bndbox").find("ymin").text)
            xmax = int(_object.find("bndbox").find("xmax").text)
            ymax = int(_object.find("bndbox").find("ymax").text)
            w = xmax - xmin
            h = ymax - ymin
            if w < 0 or h < 0:
                print(
                    "WARNING! Find error data in file {}! Box w and "
                    "h should > 0. Pass this box annotation.".format(xml_name)
                )
                continue
            coco_box = [max(xmin, 0), max(ymin, 0), min(w, width), min(h, height)]
            ann = {
                "image_id": idx + 1,  # 此处的图片序号，从1开始计算，该bbox所在的图片序号
                "bbox": coco_box,
                "category_id": cat_id,
                "iscrowd": 0,
                "id": ann_id,  # ann_id指的是bbox的序号
                "area": coco_box[2] * coco_box[3],
            }
            annotations.append(ann)  # annotation是一个list，包含了所有bbox的相关信息
            ann_id += 1

    coco_dict = {
        "images": image_info,
        "categories": categories,  # categories，从1开始计算
        "annotations": annotations,
    }

    print("Done (t={:0.2f}s)".format(time.time() - tic))

    return coco_dict


xml_folder = r"D:\documents\Learn\_Code\All_dataset\E\Annotations"
class_name = ['dog',
            ]

coco_dict = xml_to_coco(xml_folder, class_name)

"""
保存为json文件
"""
json_file = r"D:\documents\Learn\DeepLearning_Code\All_dataset\ELinage_All\COCOJson\EL_All.json"
json_fp = open(json_file, 'w')
json_str = json.dumps(coco_dict)
json_fp.write(json_str)
json_fp.close()

2、一个类的coco转VOC格式（xml）

用于一个类的情况

只需要修改6个地方：

1. CKimg_dir：生成图片保存的路径

2. CKanno_dir：生成标注文件保存的路径

3. dataTypes：annotations名称，不带后缀

4. base_dir：存放转换后的图片和标注的文件夹

5. origin_image_dir：原始的coco的图像存放位置

6. origin_anno_dir：原始的coco的标注存放位置

'''
把coco数据集合的所有标注转换到voc格式，不改变图片命名方式，
注意，原来有一些图片是黑白照片，检测出不是 RGB 图像，这样的图像不会被放到新的文件夹中
'''
from pycocotools.coco import COCO
import os, cv2, shutil
from lxml import etree, objectify
from tqdm import tqdm
from PIL import Image

# 1、生成图片保存的路径
CKimg_dir = r'faster_rcnn\balloon\VOC\images'
# 2、生成标注文件保存的路径
CKanno_dir = r'faster_rcnn\balloon\VOC\annotations'


# 若模型保存文件夹不存在，创建模型保存文件夹，若存在，删除重建
def mkr(path):
    if os.path.exists(path):
        shutil.rmtree(path)
        os.mkdir(path)
    else:
        os.mkdir(path)


def save_annotations(filename, objs, filepath):
    annopath = CKanno_dir + "/" + filename[:-3] + "xml"  # 生成的xml文件保存路径
    dst_path = CKimg_dir + "/" + filename
    img_path = filepath
    img = cv2.imread(img_path)
    im = Image.open(img_path)
    # if im.mode != "RGB":
    #     print(filename + " not a RGB image")
    #     im.close()
    #     return
    im.close()
    shutil.copy(img_path, dst_path)  # 把原始图像复制到目标文件夹
    E = objectify.ElementMaker(annotate=False)
    anno_tree = E.annotation(
        E.folder('1'),
        E.filename(filename),
        E.source(
            E.database('CKdemo'),
            E.annotation('VOC'),
            E.image('CK')
        ),
        E.size(
            E.width(img.shape[1]),
            E.height(img.shape[0]),
            E.depth(img.shape[2])
        ),
        E.segmented(0)
    )
    for obj in objs:
        E2 = objectify.ElementMaker(annotate=False)
        anno_tree2 = E2.object(
            E.name(obj[0]),
            E.pose(),
            E.truncated("0"),
            E.difficult(0),
            E.bndbox(
                E.xmin(obj[2]),
                E.ymin(obj[3]),
                E.xmax(obj[4]),
                E.ymax(obj[5])
            )
        )
        anno_tree.append(anno_tree2)
    etree.ElementTree(anno_tree).write(annopath, pretty_print=True)


def showbycv(coco,  img, classes, origin_image_dir, verbose=False):
    filename = img['file_name']
    filepath = os.path.join(origin_image_dir,  filename)
    I = cv2.imread(filepath)
    annIds = coco.getAnnIds(imgIds=img['id'], iscrowd=None)
    anns = coco.loadAnns(annIds)
    objs = []
    for ann in anns:
        name = classes[ann['category_id']]
        if 'bbox' in ann:
            bbox = ann['bbox']
            xmin = (int)(bbox[0])
            ymin = (int)(bbox[1])
            xmax = (int)(bbox[2] + bbox[0])
            ymax = (int)(bbox[3] + bbox[1])
            obj = [name, 1.0, xmin, ymin, xmax, ymax]
            objs.append(obj)
            if verbose:
                cv2.rectangle(I, (xmin, ymin), (xmax, ymax), (255, 0, 0))
                cv2.putText(I, name, (xmin, ymin), 3, 1, (0, 0, 255))
    save_annotations(filename, objs, filepath)
    if verbose:
        cv2.imshow("img", I)
        cv2.waitKey(0)

def catid2name(coco):  # 将名字和id号建立一个字典
    classes = dict()
    for cat in coco.dataset['categories']:
        classes[cat['id']] = cat['name']
        # print(str(cat['id'])+":"+cat['name'])
    return classes


def get_CK5(origin_anno_dir, origin_image_dir, verbose=False):
    dataTypes = ['annotation_coco']       # 3、根据你要转化的是训练集还是验证集（annotations名称）
    for dataType in dataTypes:
        annFile = '{}.json'.format(dataType)
        annpath = os.path.join(origin_anno_dir, annFile)
        coco = COCO(annpath)
        classes = catid2name(coco)
        imgIds = coco.getImgIds()
        # imgIds=imgIds[0:1000]#测试用，抽取10张图片，看下存储效果
        for imgId in tqdm(imgIds):
            img = coco.loadImgs(imgId)[0]
            #showbycv(coco, dataType, img, classes, origin_image_dir, verbose=False)
            showbycv(coco,  img, classes, origin_image_dir, verbose=False)


def main():
    base_dir = r'D:\documents\faster_rcnn\balloon\VOC'  # step1 这里是一个新的文件夹，存放转换后的图片和标注
    image_dir = os.path.join(base_dir, 'images')  # 在上述文件夹中生成images，annotations两个子文件夹
    anno_dir = os.path.join(base_dir, 'annotations')
    mkr(image_dir)
    mkr(anno_dir)
    # 4、
    origin_image_dir = 'train'  # step 2原始的coco的图像存放位置
    # 5、
    origin_anno_dir = 'train'  # step 3 原始的coco的标注存放位置
    print(origin_anno_dir)
    verbose = True  # 是否需要看下标记是否正确的开关标记，若是true,就会把标记展示到图片上
    get_CK5(origin_anno_dir, origin_image_dir, verbose)


if __name__ == "__main__":
    main()

2、多个类的coco转VOC格式（xml）

待补充