ImageNet file xml format to Darknet text format. Works well with directories and subdirectories.
sudo pip install -r requirements.txtpython xmltotxt.py -c cls.txt -xml xml -out out-xml -c, -outInput xml file.
<annotation>
<filename>image-0000016.jpg</filename>
<size>
<width>1920</width>
<height>1080</height>
</size>
<object>
<name>sedan</name>
<bndbox>
<xmin>75</xmin>
<ymin>190</ymin>
<xmax>125</xmax>
<ymax>210</ymax>
</bndbox>
</object>
</annotation>Output text file.
5 0.052083 0.185185 0.026042 0.018519
I used Darknet for real-time object detection and classification. Sometimes you need to collect your own training dataset for train your model. I collected training dataset images and fine awesome tool for labeling images. But it generates xml files. So I needed to implement tool which translates from ImageNet xml format to Darknet text format. Also compatible with latest YOLOv5 by Ultralytics.