This repo contains scripts for object detection task. Some tutorials written in Bahasa Indonesia available in my blog.

You can access it in folder /scripts. When you want to use it, you should move it to the parent folder.

1. xml_to_csv.py: convert PASCAL VOC xml format to csv
2. generate_labelmap.py: generate label_map.pbtxt
3. generate_tfrecord.py: generate tfrecord file
4. normalizer.py: if your subfolder format is grouped by classes. this script merge two same image with different xml files
5. folder_check.py: will raise an error if there are folder that contains different number of image and xml files
6. Explorer.ipynb: to analyze the dataset

1. Detectron.ipynb: A simple tutorial of object detection using Facebook Detectron2 Framework

I combine and modify Raccoon and Kangaroo dataset by Experincor. Datasets are available here and here. Thank You!