1. 前言

1.1 COCO数据集

COCO的全称是Common Objects in COntext，是微软团队提供的一个可以用来进行图像识别的数据集。MS COCO数据集中的图像分为训练、验证和测试集。COCO通过在Flickr上搜索80个对象类别和各种场景类型来收集图像，其使用了亚马逊的Mechanical Turk（AMT）来收集数据。COCO数据集现在有3种标注类型：object instances（目标实例）, object keypoints（目标上的关键点）, and image captions（看图说话），使用JSON文件存储。

1.2 COCO数据基本结构

这3种类型共享下面所列的基本类型，包括info、image、license，而annotation类型则呈现出了多态，会根据不同的任务具有不同的数据标注形式。

{ 
"info" : info,
"images" : [image], 
"annotations" : [annotation],
"licenses" : [license],
}
info{ 
"year" : int,
"version" : str,
"description" : str,
"contributor" : str,
"url" : str,
"date_created" : datetime,
}
image{ 
"id" : int,
"width" : int,
"height" : int,
"file_name" : str,
"license" : int,
"flickr_url" : str,
"coco_url" : str,
"date_captured" : datetime,
}
license{ 
"id" : int,
"name" : str,
"url" : str,
}

除了Annotation数据之外的数据类型举例如下：
1）info类型，比如一个info类型的实例：

"info":{ 
    "description":"This is stable 1.0 version of the 2014 MS COCO dataset.",
    "url":"http:\/\/mscoco.org",
    "version":"1.0","year":2014,
    "contributor":"Microsoft COCO group",
    "date_created":"2015-01-27 09:11:52.357475"
}

2）Images类型，Images是包含多个image实例的数组，对于一个image类型的实例：

{ 
    "license":3,
    "file_name":"COCO_val2014_000000391895.jpg",
    "coco_url":"http:\/\/mscoco.org\/images\/391895",
    "height":360,"width":640,"date_captured":"2013-11-14 11:18:45",
    "flickr_url":"http:\/\/farm9.staticflickr.com\/8186\/8119368305_4e622c8349_z.jpg",
    "id":391895
}

3）licenses类型，licenses是包含多个license实例的数组，对于一个license类型的实例：

{ 
    "url":"http:\/\/creativecommons.org\/licenses\/by-nc-sa\/2.0\/",
    "id":1,
    "name":"Attribution-NonCommercial-ShareAlike License"
}

1.3 Object Instance 类型的标注格式

1）整体JSON文件格式
Object Instance这种格式的文件从头至尾按照顺序分为以下段落：

{ 
    "info": info,
    "licenses": [license],
    "images": [image],
    "annotations": [annotation],
    "categories": [category]
}

是的，你打开这两个文件，虽然内容很多，但从文件开始到结尾按照顺序就是这5段。其中，info、licenses、images这三个结构体/类型在上一节中已经说了，在不同的JSON文件中这三个类型是一样的，定义是共享的。不共享的是annotation和category这两种结构体，他们在不同类型的JSON文件中是不一样的。

PS，mages数组、annotations数组、categories数组的元素数量是相等的，等于图片的数量。

2）annotations字段
annotations字段是包含多个annotation实例的一个数组，annotation类型本身又包含了一系列的字段，如这个目标的category id和segmentation mask。segmentation格式取决于这个实例是一个单个的对象（即iscrowd=0，将使用polygons格式）还是一组对象（即iscrowd=1，将使用RLE格式）。如下所示：

annotation{ 
    "id": int,
    "image_id": int,
    "category_id": int,
    "segmentation": RLE or [polygon],
    "area": float,
    "bbox": [x,y,width,height],
    "iscrowd": 0 or 1,
}

注意，单个的对象（iscrowd=0)可能需要多个polygon来表示，比如这个对象在图像中被挡住了。而iscrowd=1时（将标注一组对象，比如一群人）的segmentation使用的就是RLE格式。

另外，每个对象（不管是iscrowd=0还是iscrowd=1）都会有一个矩形框bbox ，矩形框左上角的坐标和矩形框的长宽会以数组的形式提供，数组第一个元素就是左上角的横坐标值。

其中，area是框的面积（area of encoded masks）。

3）categories字段
annotation结构中的categories字段存储的是当前对象所属的category的id，以及所属的supercategory的name。
categories是一个包含多个category实例的数组，而category结构体描述如下：

{ 
    "id": int,
    "name": str,
    "supercategory": str,
}

从instances_val2017.json文件中摘出的2个category实例如下所示：

{ 
    "supercategory": "person",
    "id": 1,
    "name": "person"
},
{ 
    "supercategory": "vehicle",
    "id": 2,
    "name": "bicycle"
},
......

2. 转换脚本

# -*- coding=utf-8 -*-
#!/usr/bin/python
import sys
import os
import shutil
import numpy as np
import json
import xml.etree.ElementTree as ET
# 检测框的ID起始值
START_BOUNDING_BOX_ID = 1
# 类别列表无必要预先创建，程序中会根据所有图像中包含的ID来创建并更新
PRE_DEFINE_CATEGORIES = { }
# If necessary, pre-define category and its id
# PRE_DEFINE_CATEGORIES = {"aeroplane": 1, "bicycle": 2, "bird": 3, "boat": 4,
                         # "bottle":5, "bus": 6, "car": 7, "cat": 8, "chair": 9,
                         # "cow": 10, "diningtable": 11, "dog": 12, "horse": 13,
                         # "motorbike": 14, "person": 15, "pottedplant": 16,
                         # "sheep": 17, "sofa": 18, "train": 19, "tvmonitor": 20}
def get(root, name):
    vars = root.findall(name)
    return vars
def get_and_check(root, name, length):
    vars = root.findall(name)
    if len(vars) == 0:
        raise NotImplementedError('Can not find %s in %s.'%(name, root.tag))
    if length > 0 and len(vars) != length:
        raise NotImplementedError('The size of %s is supposed to be %d, but is %d.'%(name, length, len(vars)))
    if length == 1:
        vars = vars[0]
    return vars
# 得到图片唯一标识号
def get_filename_as_int(filename):
    try:
        filename = os.path.splitext(filename)[0]
        return int(filename)
    except:
        raise NotImplementedError('Filename %s is supposed to be an integer.'%(filename))
def convert(xml_list, xml_dir, json_file):
    ''' :param xml_list: 需要转换的XML文件列表 :param xml_dir: XML的存储文件夹 :param json_file: 导出json文件的路径 :return: None '''
    list_fp = xml_list
    # 标注基本结构
    json_dict = { "images":[],
                 "type": "instances",
                 "annotations": [],
                 "categories": []}
    categories = PRE_DEFINE_CATEGORIES
    bnd_id = START_BOUNDING_BOX_ID
    for line in list_fp:
        line = line.strip()
        print("buddy~ Processing {}".format(line))
        # 解析XML
        xml_f = os.path.join(xml_dir, line)
        tree = ET.parse(xml_f)
        root = tree.getroot()
        path = get(root, 'path')
        # 取出图片名字
        if len(path) == 1:
            filename = os.path.basename(path[0].text)
        elif len(path) == 0:
            filename = get_and_check(root, 'filename', 1).text
        else:
            raise NotImplementedError('%d paths found in %s'%(len(path), line))
        ## The filename must be a number
        image_id = get_filename_as_int(filename)  # 图片ID
        size = get_and_check(root, 'size', 1)
        # 图片的基本信息
        width = int(get_and_check(size, 'width', 1).text)
        height = int(get_and_check(size, 'height', 1).text)
        image = { 'file_name': filename,
                 'height': height,
                 'width': width,
                 'id':image_id}
        json_dict['images'].append(image)
        ## Cruuently we do not support segmentation
        # segmented = get_and_check(root, 'segmented', 1).text
        # assert segmented == '0'
        # 处理每个标注的检测框
        for obj in get(root, 'object'):
            # 取出检测框类别名称
            category = get_and_check(obj, 'name', 1).text
            # 更新类别ID字典
            if category not in categories:
                new_id = len(categories)
                categories[category] = new_id
            category_id = categories[category]
            bndbox = get_and_check(obj, 'bndbox', 1)
            xmin = int(get_and_check(bndbox, 'xmin', 1).text) - 1
            ymin = int(get_and_check(bndbox, 'ymin', 1).text) - 1
            xmax = int(get_and_check(bndbox, 'xmax', 1).text)
            ymax = int(get_and_check(bndbox, 'ymax', 1).text)
            assert(xmax > xmin)
            assert(ymax > ymin)
            o_width = abs(xmax - xmin)
            o_height = abs(ymax - ymin)
            annotation = dict()
            annotation['area'] = o_width*o_height
            annotation['iscrowd'] = 0
            annotation['image_id'] = image_id
            annotation['bbox'] = [xmin, ymin, o_width, o_height]
            annotation['category_id'] = category_id
            annotation['id'] = bnd_id
            annotation['ignore'] = 0
            # 设置分割数据，点的顺序为逆时针方向
            annotation['segmentation'] = [[xmin,ymin,xmin,ymax,xmax,ymax,xmax,ymin]]
            json_dict['annotations'].append(annotation)
            bnd_id = bnd_id + 1
    # 写入类别ID字典
    for cate, cid in categories.items():
        cat = { 'supercategory': 'none', 'id': cid, 'name': cate}
        json_dict['categories'].append(cat)
    # 导出到json
    json_fp = open(json_file, 'w')
    json_str = json.dumps(json_dict)
    json_fp.write(json_str)
    json_fp.close()
if __name__ == '__main__':
    root_path = os.getcwd()
    xml_dir = os.path.join(root_path, 'Annotations')
    xml_labels = os.listdir(os.path.join(root_path, 'Annotations'))
    np.random.shuffle(xml_labels)
    split_point = int(len(xml_labels)/10)
    # validation data
    xml_list = xml_labels[0:split_point]
    json_file = './instances_val2014.json'
    convert(xml_list, xml_dir, json_file)
    for xml_file in xml_list:
        img_name = xml_file[:-4] + '.jpg'
        shutil.copy(os.path.join(root_path, 'JPEGImages', img_name),
                    os.path.join(root_path, 'val2014', img_name))
    # train data
    xml_list = xml_labels[split_point:]
    json_file = './instances_train2014.json'
    convert(xml_list, xml_dir, json_file)
    for xml_file in xml_list:
        img_name = xml_file[:-4] + '.jpg'
        shutil.copy(os.path.join(root_path, 'JPEGImages', img_name),
                    os.path.join(root_path, 'train2014', img_name))