Manga109
收藏资源简介:
Manga109是由东京大学创建的大型漫画图像数据集,包含109部由专业漫画家绘制的漫画,总计21142页。该数据集通过与版权方合作,解决了版权问题,使得研究者可以自由使用高质量的漫画图像进行研究。数据集涵盖多种漫画类型,适用于漫画图像处理、检索和本地化等研究领域,旨在提升漫画搜索的直观性、效率和乐趣。
Manga109 is a large-scale comic image dataset developed by The University of Tokyo. It consists of 109 comics created by professional cartoonists, totaling 21,142 pages. Through collaboration with copyright holders, the copyright issues associated with this dataset have been resolved, allowing researchers to freely utilize high-quality comic images for their research. Covering a wide range of comic genres, this dataset is applicable to research fields including comic image processing, retrieval and localization, aiming to enhance the intuitiveness, efficiency and enjoyment of comic searching.

- Manga109数据集首次发表,由东京大学和日本国立情报学研究所联合发布,旨在为漫画图像分析提供一个标准化的数据集。
- Manga109数据集首次应用于学术研究,特别是在计算机视觉和自然语言处理领域,推动了漫画内容理解和分析技术的发展。
- Manga109数据集的扩展版本发布,增加了更多的漫画样本和详细的标注信息,进一步提升了其在学术界的影响力。
- Manga109数据集被广泛应用于多个国际会议和竞赛中,成为评估和比较漫画分析算法性能的标准数据集之一。
- 1Manga109: A Large-scale Dataset of Manga with 109 TitlesUniversity of Tokyo · 2017年
- 2Manga OCR: A New Dataset and Baseline for Text Recognition in MangaUniversity of Tokyo · 2021年
- 3Manga Character Recognition Using Deep Learning TechniquesKyushu Institute of Technology · 2020年
- 4Exploring the Use of Manga109 Dataset for Scene Understanding in MangaUniversity of Tokyo · 2019年
- 5Manga109-s: A Dataset for Scene Understanding in MangaUniversity of Tokyo · 2022年



