A tool to convert OpenImages dataset format to TensorFlow-friendly format;
- ➰ Convert
.txt
to.xml
(VOC) format - 🚀 Make directory structure suitable to be used in TensorFlow custom object training
- 👩💻 Use presented Code Snippets for other organizing operations
string_replacer
xml_replacer
file_renamer
Dataset
|___ validation
| |___ <object_name>
| |___ Label
| | |___ <validation_label_file_name>.txt
| | |___ ....
| |___ <validation_image_file_name>.jpg
| |___ ....
|
|___ test
| |___ <object_name>
| |___ Label
| | |___ <validation_label_file_name>.txt
| | |___ ....
| |___ <validation_image_file_name>.jpg
| |___ ....
|
|___ train
|___ <object_name>
|___ Label
| |___ <validation_label_file_name>.txt
| |___ ....
|___ <validation_image_file_name>.jpg
|___ ....
📢 validation, test, train and Label are required fix keywords
images
|___ test
| |___ <object_name>_test_<id>.xml
| |___ <object_name>_test_<id>.jpg
| |___ ....
|
|___ train
|___ <object_name>_train_<id>.xml
|___ <object_name>_train_<id>.jpg
|___ ....
- Clone this repository
- Organize your data to be like the required
- Open src folder in CMD
- Run:
src> python script.py -i <INPUT_PATH> -o <OUTPUT_PATH>
👮♀️ This command will add validation set to training set folder, if you want to disable this behavior then run:
src> python script.py -i <INPUT_PATH> -o <OUTPUT_PATH> -v
👀 To see running options, run:
src> python script.py -h