数据集写txt

PA100k数据集写txt

import json
import os
# 文件

def txt2list(filename):
    f = open(filename)
    lines = f.readlines()
    file_name = []
    for index, line in enumerate(lines):
        if index == 0:
            continue
        file_name.append(line.split(".")[0].split("'")[1])  # 得到只有文件名的列表
    f.close()
    return file_name

def label2list(filename):
    f = open(filename)
    lines = f.readlines()
    ll = []
    for index, line in enumerate(lines):
        if index == 0:
            continue
        ll.append(line.strip('\n'))
    f.close()
    return ll
def writetxt(path,txt):
    with open(path,"w+") as f:
        f.write(txt)


if __name__=="__main__":
    test_img= "test_images_name.txt"
    test_img_list=txt2list(test_img)
    test_label="test_label.txt"
    test_label_list=label2list(test_label)
    path="./test_label"
    if os.path.exists(path) is False:
        os.mkdir(path)
    for i,j in zip(test_img_list,test_label_list):
        i_name=i+".txt"
        name=os.path.join(path,i_name)
        txt=i+".jpg"+" "+j.replace(","," ")
        writetxt(name,txt)

分析

import os
from pathlib import Path
import numpy as np
import matplotlib.pyplot as plt
plt.rcParams['font.sans-serif'] = 'SimHei'
def getTxtAndNum(label_dir,class_num):
    assert Path(label_dir).is_dir(), "label_dir is not exist"
    txts = os.listdir(label_dir)
    class_list = [0] * class_num  # 初始化[0, 0, 0]  存放所有txt文件中每一个类别的数量
    for txt in txts: # 遍历每一个txt
        with open(os.path.join(label_dir, txt), 'r') as f: # 打开当前txt文件 读取每一行信息 存放在lines中
            lines = f.readlines()
        for line in lines: #   line = [name 0 1 1 1]
            label = (line.split()[1:])  # 得到当前行的class信息
            for la in range(len(label)):
                if int(label[la])==1:
                    class_list[la]+=1
    data = np.array(class_list)

    return data
def draw_class_distribution(data):
    x=["0:Female","1:AgeOve r60","2:Age18-60","3:AgeLess18","4:Front","5:Side","6:Back","7:Hat","8:Glasses","9:HandBag","10:ShoulderBag","11:Backpack","12:HoLd0bjectsInFront" ,"13:ShortSLeeve" ,"14:LongSleeve","15:UpperStride","16:UpperLogo","17:UpperPlaid","18:UpperSplice" ,"19:LowerStripe ","20:LowerPattern","21:LongCoat","22:Trousers","23:Shorts","24:Skirt&Dress","25:boots"]
    print(len(x))
    fig = plt.figure(figsize=(40, 10))  # 画布大小和像素密度
    plt.bar(x, data, width=0.5, align="center")#柱状图
    # plt.plot(x, data)#折线图
    for a, b, i in zip(x, data, range(len(x))): # zip 函数
        plt.text(a, b + 0.01, "%d" % int(data[i]), ha='center', fontsize=15, color="r")  # plt.text 函数
    plt.xticks(fontsize=10)
    plt.yticks(fontsize=10)
    plt.xlabel('类别', fontsize=16)
    plt.ylabel('数量', fontsize=16)
    plt.title('PA100K测试集分布', fontsize=16)
    plt.show()
    # 保存到本地
    # plt.savefig("")


if __name__ == '__main__':
    label_dir = "./test_label"
    data = getTxtAndNum(label_dir, 26)
    print(data)
    draw_class_distribution(data)