Create copy/paste synthetic images for object detection and instance segmentation.

This repository is a modified (and extended) version of debidatta/syndata-generation and a-nau/synthetic-dataset-generation The augmentation code is not changed. The code was made more modular All credits to the original authors (also see Citation).

Clone the repository

git clone https://github.com/JureHudoklin/CopyPaste_DatasetGenerator.git

Create a virtual environment (optional) and activate it

virtualenv -p python3.8 copy_paste_env
source copy_paste_env/bin/activate

Install python dependencies

pip install -r requirements.txt

To generate a dataset two json annotation files are required.

1. A json file containing the annotations of the objects
2. A json file containing the annotations of the background images

• Objects json files should have the following information: List of objects:

  [
    {
      "name": "example_cat", # name-of-the-category
      "supercategory": "example_super_cat", #name-of-the-super-category
      "file_name": "my_img_0.jpg", #name-of-the-image-file
      "img_path": "home/my_home/images/my_img_0.jpg", #absolute-path-to-the-image-file
    },
    ...
  ]

• Background json files should have the following information: List of background images:

  [
    {
      "file_name": "my_img_0.jpg", #name-of-the-image-file
      "img_path": "home/my_home/images/my_img_0.jpg", #absolute-path-to-the-image-file
    },
    ...
  ]

You can configure the dataset generation by editing the config.py file inside the configs folder. All options are explained in the config.py file. MAKE SURE TO SET PATHS TO ANNOTATION FILES AND OUTPUT FOLDER INSIDE THE CONFIG FILE.

If you provide a load_path when creating a new dataset, the config file will be loaded from the provided path.

from configs.config import Config

cfg = Config(load_path = "path/to/config/file/filename.json"")

You can also save a config by running the following command:

from configs.config import Config

cfg = Config()
config.save(path)

To generate a dataset you can run generate_dataset.py:

python3 generate_dataset.py

If you use this code for scientific research, please consider citing the following two works on which this repository is based on.

The original work, including the code on which this repository is built. Thanks a lot to the authors for providing their code!

@InProceedings{Dwibedi_2017_ICCV,
author = {Dwibedi, Debidatta and Misra, Ishan and Hebert, Martial},
title = {Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detection},
booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
month = {Oct},
year = {2017}
}

• Paper: arxiv and ICCV 2017

Our work for which this repository was developed.

@inproceedings{naumannScrapeCutPasteLearn2022,
  title = {Scrape, Cut, Paste and Learn: Automated Dataset Generation Applied to Parcel Logistics},
  booktitle = {{{IEEE Conference}} on {{Machine Learning}} and Applications} ({{ICMLA}})},
  author = {Naumann, Alexander and Hertlein, Felix and Zhou, Benchun and Dörr, Laura and Furmans, Kai},
  date = {2022},
}