LICENSE Agreement of Manga Segmentation Annotations version 1.0
-----------------------------------------------------------------
Dated: 2024-06-12
License document prepared by  Minshan Xie

All segmentation annotations are owned and copyrighted by Minshan Xie 
and Tien-Tsin Wong.


Purpose of Manga Segmentation Annotations
-----------------------------------------------------------------
The annotation aims to generate training data for deep learning-based 
instance segmentation models of manga images. These annotations allow 
us to identify individual components within a black-and-white manga 
page. Unfortunately, the current bounding box annotations in the 
Manga109 dataset, the largest-scale black-and-white manga dataset, 
are insufficient for many manga-related applications, such as manga 
inpainting, localization, and colorization. Hence, we have developed 
segmentation annotations for the Manga109 dataset.

We have annotated six types of manga components at the instance level 
within the Manga109 dataset, i.e. frame, text, face, body, balloon, 
onomatopoeia. With these annotations, we can train deep-learning models 
for fast manga segmentation. After segmenting these components, we can 
seamlessly adapt manga to other media formats, such as anime, video 
games, or interactive storytelling.


Permission to Use
-----------------------------------------------------------------
You are automatically granted permission to use the images in the 
segmentation annotations for both academic and commercial purposes. 
However, please ensure that you include the following credit in any 
form of publication, reproduction, redistribution, or derivatives of 
the images:

- Minshan Xie, Hanyuan Liu, Jian Lin, Chengze Li, and Tien-Tsin Wong,
  "Advancing Manga Analysis: Comprehensive Segmentation Annotations 
  for the Manga109 Dataset," arXiv preprint, 2024.

If you find the dataset useful, we kindly request that you cite our 
publication in your work or publication.