目标检测xml标注转txt标注

目标检测xml标注转txt标注

FAIR1M数据集的原始标注格式为xml标注，具体为：

而DOTA数据集标注为txt文件，格式如下：

我想用DOTA数据集的格式去训练FAIR1M数据集，代码如下：

import os
import sys
import xml.etree.ElementTree as ET
import glob

def xml_to_txt(indir, outdir):

    # os.chdir(indir)      #  切换到给定目录下        
    annotations = os.listdir(indir)

    for i, file in enumerate(annotations):

        file_save = file.split('.')[0]+'.txt'
        file_txt=os.path.join(outdir, file_save)
        f_w = open(file_txt, mode='w+', encoding='UTF-8')

        # actual parsing
        in_file = open(indir + '/' + file)
        tree = ET.parse(in_file)
        root = tree.getroot()

        for obj in root.iter('object'):
            current = list()
            name = obj.find('possibleresult').find('name').text
            name = name.replace(" ", "-")  #FAIR1M的类别标注会有空格，用-代替

            xmlbox = obj.find('points')
            points = xmlbox.findall('point')

            poly_list = []
            for point in points:
                point = point.text
                point = point.split(",")
                poly_list = poly_list + point
            f_w.write(poly_list[0] + ' ' + poly_list[1] + ' '+ poly_list[2] + ' ' + poly_list[3] + ' ' +
                       poly_list[4] + ' '+ poly_list[5] + ' ' + poly_list[6] + ' ' + poly_list[7] + ' ' + name + ' ' + '0' + '\n')
        f_w.close()

indir='train/labelXml'   #xml目录
outdir='train/labelTxt'  #txt目录

if __name__ == "__main__":
    xml_to_txt(indir, outdir)

另外给出FAIR1M的类别名称统计代码：

import os
import os.path
import numpy as np
import cv2
import matplotlib.pyplot as plt

train_list = os.listdir("/workspace/FAIR1M2/train/labelTxt/")
base_path = "/workspace/FAIR1M2/train/labelTxt/"

total_dict = []

for file in train_list:
    with open(base_path + file) as f:
        s = f.readlines()
        for si in s:
            bbox_info = si.split()
            if bbox_info[8] in total_dict:
                continue
            else:
                total_dict.append(bbox_info[8])
print(total_dict)
print(len(total_dict))