Manga FaceNet: Face Detection in Manga based on Deep Neural Network

Wei-Ta Chu and Wei-Wei Li

Multimedia Computing Laboratory
Dept. of Computer Science and Information Engineering
National Chung Cheng University

1. Introduction

Among various elements of manga, character's face plays one of the most important role in access and retrieval. We propose a deep-based method to do manga face detection, which is a challenging but relatively unexplored topic. Given a manga page, we first find candidate regions based on the selective search scheme. A deep neural network is then proposed to detect manga faces of various appearance. We evaluate the proposed method based on a large-scale benchmark, and show performance comparison and convincing evaluation results that have rarely done before.

2. Manga Face Dataset

We selected part of the Manga109 dataset, and manually label positions of manga faces. Please directly go to http://www.manga109.org/en/ to download the manga pages.

IMPORTANT!

    Please notice that, when you download files from the Manga109 dataset, you will find each .jpg file contains two manga pages put side-by-side. In order to take one manga page as one unit, we evenly divide each .jpg file vertically into two files and rename them. The selected pages and ground truth information provided below are based on the manipulated files. The mapping between original Manga109 files and the manipulated files is listed in the mapping.csv provided below. In the mapping file, you will find, for example,

    This means, in the "RisingGirl" manga title, the original file name 3.jpg is evenly and vertically divided into two pages. The left half is named as "RisingGirl_007.jpg" and the right half is named as "RisingGirl_006.jpg". Sorry that we cannot directly provide the manipulated files here because the limitation of dataset usage. For users who are interested in using the ground truth provided below, please divide the original files into two pages and follow the naming shown in the mapping.csv. (Sorry again for this awkward naming.)

3. Citation

Please cite our work if you utilize this dataset.

Wei-Ta Chu and Wei-Wei Li, "Manga Face Detection based on Deep Neural Networks Fusing Global and Local Information," Pattern Recognition, vol. 86, pp. 62-72, 2019.
Wei-Ta Chu and Wei-Wei Li, "Manga FaceNet: Face Detection in Manga based on Deep Neural Network," Proceedings of ACM International Conference on Multimedia Retrieval, pp. 412-415, 2017.

 

Any problem please contact .

[Main Page]

Last Updated: Feb. 12, 2019