Metadata-Version: 2.1
Name: make_voc_dataset
Version: 0.0.5.dev2
Summary: A quick tool to make VOC style Dataset
Home-page: https://github.com/Razin-Tailor/make_voc_dataset
Author: Razin Tailor
Author-email: razintailorpy@gmail.com
License: UNKNOWN
Project-URL: Bug Tracker, https://github.com/Razin-Tailor/make_voc_dataset/issues
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.6
Description-Content-Type: text/markdown
License-File: LICENSE

# make-voc-dataset

Majority of the current Deep Learning Frameworks like [MMDetection](https://github.com/open-mmlab/mmdetection) or [Detectron2](https://github.com/facebookresearch/detectron2) support the **VOC Formatted Data / COCO Formatted Data**

This simple tool helps convert files stored in your local machine in the VOC Formatted Directory Structure

The data should have images and their corresponding annotation_file in **xml** format.

Tools like [LabelIMG](https://github.com/tzutalin/labelImg) can be used to annotate images.

## Options

`-s / --source_path`
Provide the full path where all the images and xmls are located. Please make sure the img_names are **UNIQUE**

`-d / --dest_path`
Provide the full to save the VOC structure

`-t / --test_ratio`
Provide the ratio (between 0 and 1) that will be the test division of the dataset


