目标检测数据集制作常用脚本集合

1、VOC格式篇

这个代码的用途是将annotations下的xml文件转换成txt形式的labels。

# 第二步运行
import xml.etree.ElementTree as ET
import pickle
import os
from os import listdir, getcwd
from os.path import join
sets=[('2007', 'train'), ('2007', 'val'), ('2007', 'test')]  #替换为自己的数据集
classes = ["person","bird","cat","cow","dog","horse",
"sheep","aeroplane","bicycle","boat","bus","car","motorbike","train","bottle","chair","diningtable","pottedplant","sofa","tvmonitor"]     #修改为自己的类别

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(year, image_id):
    in_file = open('Annotations/%s.xml'%(image_id))  #将数据集放于当前目录下
    out_file = open('label/%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
        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))
        bb = convert((w,h), b)
        out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
wd = getcwd()
for year, image_set in sets:
    if not os.path.exists('labels/'):
        os.makedirs('labels/')
    image_ids = open('ImageSets/Main/%s.txt'%(image_set)).read().strip().split()
    list_file = open('%s_%s.txt'%(year, image_set), 'w')
    for image_id in image_ids:
        # list_file.write('JPEGImages/%s.jpg\n'%(image_id))
        list_file.write('%s.jpg\n'%(image_id))
        convert_annotation(year, image_id)
    list_file.close()   
# os.system("cat 2007_train.txt 2007_val.txt > train.txt")     #修改为自己的数据集用作训练

2、COCO格式篇

3、图像尺寸变换

import datetime
import io, os
from PIL import Image
import shutil
import cv2
import numpy as np
import matplotlib.pyplot as plt 

def crop_image(image, size):   # image_dir 批量处理图像文件夹 size 裁剪后的尺寸
        h, w = image.shape[0:2]
        h_no = h // size
        w_no = w // size

        for row in range(0, h_no):
            for col in range(0, w_no):
                cropped_img = image[size*row : size*(row+1), size*col : size*(col+1), : ]
        return cropped_img



inputPathRoot = 'D:\\Users\\my_SAR\\'
outputPathRoot = 'D:\\Users\\my_SAR_jpg\\'

flag_change_format = 1
# 读取图像, 转换格式, 裁剪, 加黑边(保持分辨率)
if flag_change_format:
    print(str(datetime.datetime.now()) + " 图片格式转换功能激活 ")
    counter = 0
    for roots, dirs, files in os.walk(inputPathRoot):
        for index in range(len(files)):
            formatFile = files[index].split('.')[-1]
            name = files[index].split('.')[-2]
#            print(name)
            if formatFile == 'JPG':
                oldFilePath = os.path.join(roots, files[index])
#                print(oldFilePath)
                img = cv2.imread(oldFilePath)
                newroots = roots.replace(inputPathRoot,outputPathRoot)
                if not os.path.exists(newroots):
                    os.makedirs(newroots)
                newFilePath = os.path.join(newroots, name)
                newFilePath = str(newFilePath) + '.jpg'
#                print(newFilePath)
                
                h, w, c = img.shape
                if h<=128 and w<=128:
#cv2.copyMakeBorder(img, top_size, bottom_size, left_size, right_size, cv2.BORDER_CONSTANT, value=0)
                    img = cv2.copyMakeBorder(img,64-h//2, 64-h//2, 64-w//2, 64-w//2,cv2.BORDER_CONSTANT, value=0)
                    img2 = cv2.resize(img, (128, 128), interpolation=cv2.INTER_CUBIC)
                elif h>128 and w<=128:
                    img = cv2.copyMakeBorder(img,h,128,0,0,cv2.BORDER_CONSTANT,value=0)
                    img2 = crop_image(img, 128)
                elif h<=128 and w>128:
                    img = cv2.copyMakeBorder(img, 128, w,0,0,cv2.BORDER_CONSTANT,value=0)
                    img2 = crop_image(img, 128)
                else:
                    img2 = crop_image(img, 128)
                       
                cv2.imwrite(newFilePath, img2)
                counter += 1
    print('已经转换 %s 张图' % (counter))
    print(str(datetime.datetime.now()) + " 完成 ! ! ! ! ! ! ")


flag_transform = 0
# 搜索数据中特定格式的文件, 拷贝到另一个文件夹内
if flag_transform:
    print(str(datetime.datetime.now()) + " 开始转换 ......... ")
    counter = 0
    for roots, dirs, files in os.walk(inputPathRoot):
        for index in range(len(files)):
            formatFile = files[index].split('.')[-1]
            if formatFile == 'tif':
    #        print(files)
    #        for name in files:
    #            if '.JPG' in name:
    #                print(name)
                oldFilePath = os.path.join(roots, files[index])
                
                if 'Patch_Uint8' in oldFilePath:
                    newroots = roots.replace(inputPathRoot,outputPathRoot)
                    if not os.path.exists(newroots):
                        os.makedirs(newroots)
                    newFilePath = os.path.join(outputPathRoot, files[index])
                    shutil.copyfile(oldFilePath, newFilePath)
    #            os.system("mstar2jpeg -i %s -o %s -q 95" % (oldFilePath,newFilePath))
                    counter += 1
    print('已经转换 %s 张图' % (counter))
    print(str(datetime.datetime.now()) + " 完成 ! ! ! ! ! ! ")
else :
    print('转换功能未激活')