德国交通标志识别训练集、验证集的划分，测试集读取

需求：读取csv文件的最后一列，真是底子不实在，曲线救国啊！

先读出来的是字符串的列表，不知道为何把表头也读进来了，

然后把他踢出去（出栈），

然后再转化为整型的列表。

学会把每一步输出，学会debug！

还有就是，字符串组成的列表也可以对列表中的每个字符串进行切片操作

import csv
def read_label(dir):
    with open(dir, "r") as f:
        #"D:\\dataset\\GTSRB\\GT-final_test.csv"
        reader = csv.reader(f)
        list = [i[0].split(";")[7] for i in reader]
        # print(list)
        list.pop(0)
        # print(list)
        new_numbers = [];
        for n in list:
          new_numbers.append(int(n));
        numbers = new_numbers;
        # print(numbers)
    return numbers

 德国交通标志识别训练和测试集读取

import numpy as np
import torch
import os
import csv
import random
from PIL import Image
from torch.utils.data import Dataset
random.seed(1)
tsr_label={"0": 0, "1": 1, "2": 2, "3": 3,"4": 4, "5": 5, "6": 6, "7": 7,"8": 8, "9": 9, "10": 10,"11": 11,"12": 12, "13": 13, "14": 14, "15": 15,"16": 16, "17": 17, "18": 18,
         "19": 19,"20": 20, "21": 21, "22": 22, "23": 23,"24": 24, "25": 25, "26": 26, "27": 27, "28": 28, "29": 29, "30": 30, "31": 31,"32": 32, "33": 33, "34": 34, "35": 35,
         "36": 36, "37": 37, "38": 38, "39": 39,"40": 40, "41": 41, "42": 42
         }


# class TSRDataset(Dataset):
class CatDogDataset(Dataset):

    def __init__(self,data_dir,mode="train",transform=None):
        self.label_name = {"0": 0, "1": 1, "2": 2, "3": 3,"4": 4, "5": 5, "6": 6, "7": 7,"8": 8, "9": 9, "10": 10,"11": 11,"12": 12, "13": 13, "14": 14, "15": 15,"16": 16, "17": 17, "18": 18,
         "19": 19,"20": 20, "21": 21, "22": 22, "23": 23,"24": 24, "25": 25, "26": 26, "27": 27, "28": 28, "29": 29, "30": 30, "31": 31,"32": 32, "33": 33, "34": 34, "35": 35,
         "36": 36, "37": 37, "38": 38, "39": 39,"40": 40, "41": 41, "42": 42
         }
        self.mode = mode
        self.data_info = self.get_img_info(self,data_dir)# 是个list，jpg格式的数据；分类情况
        # # data_info存储所有图片路径和标签，在DataLoader中通过index读取样本
        self.transform = transform

    def __getitem__(self, index):
        # 根据index索引返回图片、标签 todo
        path_img, label = self.data_info[index]
        # 在init里定义，用下面定义的方法，用来读取图片路径和标签
        img = Image.open(path_img).convert('RGB')     # 0~255
        # img是PIL数据类型
        if self.transform is not None:
            img = self.transform(img)   # 在这里做transform，转为tensor等等
        return img, label

    def __len__(self):
        return len(self.data_info)

    @staticmethod
    def get_img_info(self,data_dir):
        data_info = list()
        if self.mode == "train":
            for root, dirs, _ in os.walk(data_dir):
                # 遍历类别
                for sub_dir in dirs:
                    img_names = os.listdir(os.path.join(root, sub_dir))
                    img_names = list(filter(lambda x: x.endswith('.jpg'), img_names))
                    # 遍历图片
                    for i in range(len(img_names)):
                        img_name = img_names[i]
                        path_img = os.path.join(root, sub_dir, img_name)
                        label = tsr_label[sub_dir]
                        data_info.append((path_img, int(label)))
                        # path_img图片在计算机中的路径
                        # 图片的类别
        else:
            data_info = self.read_test_dataset(data_dir)
            # data_dir = r"D:\dataset\myGTSRBTest"
            # img_names = os.listdir(data_dir)
            # img_names = list(filter(lambda x: x.endswith('.jpg'), img_names))  # 列表
            # path_img_set = [os.path.join(data_dir, n) for n in img_names]
            # img_labels = self.read_label(r"D:\dataset\myTestcsv\GT-final_test.csv")
            # data_info = [(n, l) for n, l in zip(path_img_set, img_labels)]
        return data_info

    def read_test_dataset(self,dir):
        import os
        import csv
        # data_dir = r"D:\dataset\myGTSRBTest"
        img_names = os.listdir(dir)
        img_names = list(filter(lambda x: x.endswith('.jpg'), img_names))
        # print(img_names)
        csv_file = open(r"GT-final_test.csv")  # 打开csv文件
        csv_reader_lines = csv.reader(csv_file)
        dict = {}
        path_dict = {}
        # data = []# 不用这个
        for one_line in csv_reader_lines:
            # data.append(one_line[len(one_line)-1])
            dict[int(one_line[0])] = one_line[len(one_line) - 1]

        for root, dirs, _ in os.walk(dir):
            for i in range(len(img_names)):
                img_name = img_names[i]
                path_img = os.path.join(root, img_name)
                path_dict[i] = path_img
        result_list = []
        for _ in range(len(dict)):
            temp_tuple = []
            temp_tuple.append(path_dict[_])
            temp_tuple.append(int(dict[_]))
            result_list.append(tuple(temp_tuple))
        return result_list

 训练集验证集的划分

    def get_img_info(self,data_dir):
        import random
        rng_seed = 620
        # data_dir = r"D:\dataset\myGTSRB"
        # mode = "train"
        split_idx = 0.9
        random.seed(rng_seed)
        data_info = self.read_data_info(data_dir)
        random.shuffle(data_info)
        indx = int(len(data_info) * split_idx)
        if self.mode == "train":
            train = data_info[:indx]
            # print(len(train))
        else:
            train = data_info[indx:]
        return train

    def read_data_info(self,data_dir):
        import os
        data_info = list()
        for root, dirs, _ in os.walk(data_dir):
            # 遍历类别
            # label = 0
            # path_img = None
            for sub_dir in dirs:
                img_name = os.listdir(os.path.join(root, sub_dir))
                # 用于返回指定的文件夹包含的文件或文件夹的名字的列表
                # img_names = list(filter(lambda x: x.endswith('.jpg'), img_name))
                # 遍历图片

                for i in range(len(img_name)):
                    img_name1 = img_name[i]
                    path_img = os.path.join(root, sub_dir, img_name1)
                    label = tsr_label[sub_dir]
                    data_info.append((path_img, int(label)))
                    # print(data_info)
                # print(data_info)
        return data_info

 注意：不要介意那些类的命名，变量的命名，我是个小白。