Yolov3 trains its own data set to generate train txt

from os.path import join
import os
img_dir = "images"
work_dir = os.getcwd()
print (" Current working directory  : %s" % work_dir)
txt_file = open(work_dir+'/train.txt', 'w')
img_path = join(work_dir,img_dir)
print (" Picture folder  :",img_path)
image_names = os.listdir('./'+ img_dir)
for file_name in image_names:
    if file_name.endswith(".jpg"):
        txt_file.write(img_path + "/"+ file_name+"\n")
txt_file.close()

work_dir This is it py The path of the script , such as E:\workspace\classify

img_dir Is the path of the dataset image , For example, I put all the pictures in ,E:\workspace\classify\images in

I'm going to put work_dir +“images”, Got it. img_dir, If your path is different from, modify it by yourself

Then traverse img_dir All under jpg picture , If you are another picture, you put if file_name.endswith(".jpg") Make a change , Don't judge the suffix , Write the absolute path of each picture in txt in .

Pay attention to the , stay win System generated txt, Can't put linux of use , Because of the backslash .