coco数据集和voc数据集的划分与使用

一、coco2017的目录结构

但coco数据集太大了，划分至原来的四分之一（参考链接）：

# coding:utf8
__first_version_author__ = 'tylin'
__second_version_author__ = 'wfnian'
# Interface for accessing the Microsoft COCO dataset.

# Microsoft COCO Toolbox.	  version 2.0
# Data, paper, and tutorials available at:  http://mscoco.org/
# Code written by Piotr Dollar and Tsung-Yi Lin, 2014.
# Licensed under the Simplified BSD License [see bsd.txt]

import json
import time
import shutil
import os
from collections import defaultdict
import json
from pathlib import Path


class COCO:
    def __init__(self, annotation_file=None, origin_img_dir=""):
        """
        Constructor of Microsoft COCO helper class for reading and visualizing annotations.
        :param annotation_file (str): location of annotation file
        :param image_folder (str): location to the folder that hosts images.
        :return:
        """
        # load dataset
        self.origin_dir = origin_img_dir
        self.dataset, self.anns, self.cats, self.imgs = dict(), dict(), dict(), dict()  # imgToAnns　一个图片对应多个注解(mask) 一个类别对应多个图片
        self.imgToAnns, self.catToImgs = defaultdict(list), defaultdict(list)
        if not annotation_file == None:
            print('loading annotations into memory...')
            tic = time.time()
            dataset = json.load(open(annotation_file, 'r'))
            assert type(dataset) == dict, 'annotation file format {} not supported'.format(type(dataset))
            print('Done (t={:0.2f}s)'.format(time.time() - tic))
            self.dataset = dataset
            self.createIndex()

    def createIndex(self):
        # create index　　  给图片->注解,类别->图片建立索引
        print('creating index...')
        anns, cats, imgs = {}, {}, {}
        imgToAnns, catToImgs = defaultdict(list), defaultdict(list)
        if 'annotations' in self.dataset:
            for ann in self.dataset['annotations']:
                imgToAnns[ann['image_id']].append(ann)
                anns[ann['id']] = ann

        if 'images' in self.dataset:
            for img in self.dataset['images']:
                imgs[img['id']] = img

        if 'categories' in self.dataset:
            for cat in self.dataset['categories']:
                cats[cat['id']] = cat

        if 'annotations' in self.dataset and 'categories' in self.dataset:
            for ann in self.dataset['annotations']:
                catToImgs[ann['category_id']].append(ann['image_id'])

        print('index created!')

        # create class members
        self.anns = anns
        self.imgToAnns = imgToAnns
        self.catToImgs = catToImgs
        self.imgs = imgs
        self.cats = cats

    def build(self, tarDir=None, tarFile='./new.json', N=1000):

        load_json = {'images': [], 'annotations': [], 'categories': [], 'type': 'instances', "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"}, "licenses": [{"url": "http:\/\/creativecommons.org\/licenses\/by-nc-sa\/2.0\/", "id": 1, "name": "Attribution-NonCommercial-ShareAlike License"}, {"url": "http:\/\/creativecommons.org\/licenses\/by-nc\/2.0\/", "id": 2, "name": "Attribution-NonCommercial License"}, {"url": "http:\/\/creativecommons.org\/licenses\/by-nc-nd\/2.0\/",
                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              "id": 3, "name": "Attribution-NonCommercial-NoDerivs License"}, {"url": "http:\/\/creativecommons.org\/licenses\/by\/2.0\/", "id": 4, "name": "Attribution License"}, {"url": "http:\/\/creativecommons.org\/licenses\/by-sa\/2.0\/", "id": 5, "name": "Attribution-ShareAlike License"}, {"url": "http:\/\/creativecommons.org\/licenses\/by-nd\/2.0\/", "id": 6, "name": "Attribution-NoDerivs License"}, {"url": "http:\/\/flickr.com\/commons\/usage\/", "id": 7, "name": "No known copyright restrictions"}, {"url": "http:\/\/www.usa.gov\/copyright.shtml", "id": 8, "name": "United States Government Work"}]}
        if not Path(tarDir).exists():
            Path(tarDir).mkdir()

        for i in self.imgs:
            if(N == 0):
                break
            tic = time.time()
            img = self.imgs[i]
            load_json['images'].append(img)
            fname = os.path.join(tarDir, img['file_name'])
            anns = self.imgToAnns[img['id']]
            for ann in anns:
                load_json['annotations'].append(ann)
            if not os.path.exists(fname):
                shutil.copy(self.origin_dir+'/'+img['file_name'], tarDir)
            print('copy {}/{} images (t={:0.1f}s)'.format(i, N, time.time() - tic))
            N -= 1
        for i in self.cats:
            load_json['categories'].append(self.cats[i])
        with open(tarFile, 'w+') as f:
            json.dump(load_json, f, indent=4)


coco = COCO('../../datasets/coco/annotations/instances_train2017.json',
            origin_img_dir='../../datasets/coco/train2017')               # 完整的coco数据集的图片和标注的路径
coco.build('./mini_train2017', './mini_instances_train2017.json', 29568)  # 保存图片路径


coco = COCO('../../datasets/coco/annotations/instances_val2017.json',
            origin_img_dir='../../datasets/coco/val2017')                 # 完整的coco数据集的图片和标注的路径
coco.build('./mini_val2017', './mini_instances_val2017.json', 1250)       # 保存图片路径

# 在2017年数据集中,训练集118287张,验证5000张,测试集40670张.
# 118287/4 = 29568 5000/4 = 1250

划分后结构如下：

在代码的coco.yaml文件中，更改如下语句：

path:  E:\code\yolo\yolov5-2\coco\images\
# dataset root dir <PATH>/coco  #E:/dataset/coco/

train: train2017  # train images (relative to 'path') 118287 images train2017.txt
val: val2017  # train images (relative to 'path') 5000 images val2017.txt
#test: test-dev2017.txt  # 20288 of 40670 images, submit to https://competitions.codalab.org/competitions/20794

二、voc2012
1、目录结构如下：

目标检测只需要前三个文件夹。
ImageSets中除去main的文件夹都可以删掉，main中的所有txt文件也可以删掉，一般都是自己划分。
在此之前将JPEGImages文件夹改名为images。（该步非常重要，否则会找不到标签）
现在目录结构为：

在根目录（这三个文件夹的同一路径下）创建split_train_val.py，将voc划分为8:1:1（参考链接）：

# -*- coding: utf-8 -*-
"""
Author:smile
Date:2022/09/11 10：00
顺序：脚本A1
简介：分训练集、验证集和测试集，按照 8：1：1 的比例来分，训练集8，验证集1，测试集1

"""
import os
import random
import argparse

parser = argparse.ArgumentParser()
# xml文件的地址，根据自己的数据进行修改 xml一般存放在Annotations下
parser.add_argument('--xml_path', default='E:/dataset/VOC2012ok/Annotations/', type=str, help='input xml label path')
# 数据集的划分，地址选择自己数据下的ImageSets/Main
parser.add_argument('--txt_path', default='E:/dataset/VOC2012ok/ImageSets/Main/', type=str, help='output txt label path')
opt = parser.parse_args()

train_percent = 0.8  # 训练集所占比例
val_percent = 0.1  # 验证集所占比例
test_persent = 0.1  # 测试集所占比例

xmlfilepath = opt.xml_path
txtsavepath = opt.txt_path
total_xml = os.listdir(xmlfilepath)

if not os.path.exists(txtsavepath):
    os.makedirs(txtsavepath)

num = len(total_xml)
list = list(range(num))

t_train = int(num * train_percent)
t_val = int(num * val_percent)

train = random.sample(list, t_train)
num1 = len(train)
for i in range(num1):
    list.remove(train[i])

val_test = [i for i in list if not i in train]
val = random.sample(val_test, t_val)
num2 = len(val)
for i in range(num2):
    list.remove(val[i])

file_train = open(txtsavepath + '/train.txt', 'w')
file_val = open(txtsavepath + '/val.txt', 'w')
file_test = open(txtsavepath + '/test.txt', 'w')

for i in train:
    name = total_xml[i][:-4] + '\n'
    file_train.write(name)

for i in val:
    name = total_xml[i][:-4] + '\n'
    file_val.write(name)

for i in list:
    name = total_xml[i][:-4] + '\n'
    file_test.write(name)

file_train.close()
file_val.close()
file_test.close()

运行完毕后会在iImageSets-Main文件夹下生成三个txt文件。

再创建voc_label.py:

# -*- coding: utf-8 -*-
import xml.etree.ElementTree as ET
import os

sets = ['train', 'val', 'test']  # 如果你的Main文件夹没有test.txt，就删掉'test'
# classes = ["a", "b"]   # 改成自己的类别，VOC数据集有以下20类别
classes = ["aeroplane", 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog',
           'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor']  # class names
abs_path = os.getcwd()


def convert(size, box):
    dw = 1. / (size[0])
    dh = 1. / (size[1])
    x = (box[0] + box[1]) / 2.0 - 1
    y = (box[2] + box[3]) / 2.0 - 1
    w = box[1] - box[0]
    h = box[3] - box[2]
    x = x * dw
    w = w * dw
    y = y * dh
    h = h * dh
    return x, y, w, h


def convert_annotation(image_id):
    in_file = open(abs_path + '/Annotations/%s.xml' % (image_id), encoding='UTF-8')
    out_file = open(abs_path + '/labels/%s.txt' % (image_id), 'w')
    tree = ET.parse(in_file)
    root = tree.getroot()
    size = root.find('size')
    w = int(size.find('width').text)
    h = int(size.find('height').text)
    for obj in root.iter('object'):
        difficult = obj.find('difficult').text
        # difficult = obj.find('Difficult').text
        cls = obj.find('name').text
        if cls not in classes or int(difficult) == 1:
            continue
        cls_id = classes.index(cls)
        xmlbox = obj.find('bndbox')
        b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text),
             float(xmlbox.find('ymax').text))
        b1, b2, b3, b4 = b
        # 标注越界修正
        if b2 > w:
            b2 = w
        if b4 > h:
            b4 = h
        b = (b1, b2, b3, b4)
        bb = convert((w, h), b)
        out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')


for image_set in sets:
    if not os.path.exists(abs_path + '/labels/'):
        os.makedirs(abs_path + '/labels/')

    image_ids = open(abs_path + '/ImageSets/Main/%s.txt' % (image_set)).read().strip().split()
    list_file = open(abs_path + '/%s.txt' % (image_set), 'w')
    for image_id in image_ids:
        list_file.write(abs_path + '/images/%s.jpg\n' % (image_id))  # 要么自己补全路径，只写一半可能会报错
        convert_annotation(image_id)
    list_file.close()

会在根目录生成三个txt文件：

之后使用数据集用的就是根目录这三个txt文件，train.txt val.txt test.txt.
在使用时，需要修改voc.yaml文件：

path:  E:\dataset\VOC2012ok #<PATH>/VOC

train: train.txt # train images (relative to 'path')  16551 images

val: val.txt # val images (relative to 'path')  4952 images
  
test: test.txt # test images (optional)

引用：
1、
版权声明：本文为CSDN博主「zstar-_」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/qq1198768105/article/details/125645443
2、https://zhuanlan.zhihu.com/p/423898204