深度学习中生成图像列表的txt文件

1.深度学习中经常要将图像的文件名写入txt目录中

首先  ls -l *.tif >1.txt

然后，用notepad++的多行处理功能处理生成的txt文件

2.用python编程，读取一个文件夹下所有指定的文件的文件名到List，然后操作该List，按照train,test,val划分，将划分后的文件名写入文本中

import glob
import random
imgs_listFull = glob.glob("/media/mlxuan/winD/DeepLearningData/ISPRS_BENCHMARK_DATASETS/Potsdam/imageIRRG/*.tif")#返回所匹配的文件名列表
imgs_list=[i[len("/media/mlxuan/winD/DeepLearningData/ISPRS_BENCHMARK_DATASETS/Potsdam/imageIRRG/"):] for i in imgs_listFull]
train_per = 0.8
valid_per = 0.1
test_per = 0.1

random.seed(666)
random.shuffle(imgs_list)

imgs_num = len(imgs_list)
train_point = int(imgs_num * train_per)
valid_point = int(imgs_num *  valid_per)


trainList = imgs_list[0:train_point]
valList = imgs_list[train_point:train_point+valid_point]
testList = imgs_list[train_point+valid_point:imgs_num-1]

fileTrainIRRG = open("/media/mlxuan/winD/DeepLearningData/ISPRS_BENCHMARK_DATASETS/Potsdam/txtForTrain/trainIRRG.txt",'w')
fileValIRRG = open("/media/mlxuan/winD/DeepLearningData/ISPRS_BENCHMARK_DATASETS/Potsdam/txtForTrain/valIRRG.txt",'w')
fileTestIRRG = open("/media/mlxuan/winD/DeepLearningData/ISPRS_BENCHMARK_DATASETS/Potsdam/txtForTrain/testIRRG.txt",'w')


for line in trainList:
    line2 = line+' '+line.replace("IRRG","label")
    fileTrainIRRG.write(line2)
    fileTrainIRRG.write('\n')

for line in valList:
    line2 = line+' '+line.replace("IRRG","label")
    fileValIRRG.write(line)
    fileValIRRG.write('\n')
    
for line in testList:
    line2 = line+' '+line.replace("IRRG","label")
    fileTestIRRG.write(line2)
    fileTestIRRG.write('\n')    
    
    
fileTrainIRRG.close()
fileValIRRG.close()
fileTestIRRG.close()

 

 

 

 

2.不好操作的直接用python编程，比如下面程序，读取源文件中的每一行，对每一行操作后写入新的文件中

file = open("trainLable.txt",'w')

with open("train2.txt", "r") as f:
    line = f.readline()
    while line:
        line2 = line[0:-1]+' '+line.replace("RGB","label")#[0:-1]是为了去除换行键
        file.write(line2)
        line = f.readline()

file.close


常见的对一行的其他操作：从''-rw-r--r-- 1 mlxuan mlxuan 360154 10月 10 21:15 0.tiff\n'获取最后的0.tiff:
line = f.readline()
imagename = line[0:-1].split()[-1]