This is a simple python program for the four .txt file in ImageSets/Main document creating, which is used to build your train sets for Faster R-CNN.
The proportion of each part of sets should be：
test: 50% of all
trainval: 50% of all
train: 50% of trainval
val: 50% of trainval
This program is fully automatic which means it will read the content of Annotations of current dir and create needed dir or file and fill the .txt file.
这是一个帮助创建faster R-CNN 自用训练集的程序。
他会帮助你建立好在ImageSets/Main下的四个txt文件： test,trainval,train,val，并且按照规定的比例分配。