Converting the data set format

Ha, now the VHR-10 VOC converted format data set, are recorded as follows:
Total Reference Links: https://blog.csdn.net/weixin_43202256/article/details/83058545
Step1:
Rename Image: HTTPS: / /blog.csdn.net/u011574296/article/details/72956446
I chose python version of the code, but note that directly after you run this code, the name of the picture will start from 000000, needs its own little modify the code. See convert_image.py codes in RFB-Net.
step2:
the label information into xml file:
Transferable: https://blog.csdn.net/summer2day/article/details/83064727.
Changed some routing problem, see specific pycharm.
We realized the need: to convert each .txt file into xml, and there is another folder in Annotations.
Note that xml file information: https://m.imooc.com/article/27739 picture width and height, the picture target information.
step3:
generating test.txt, trainval.txt and other documents. Code links to see the total, need to be patient.
Although this is a small job, but the feeling is very useful for laying the foundation. Yesterday afternoon has been reviewed.