Home

Awesome

Bandai-Namco-Research-Motiondataset

This repository provides motion datasets collected by Bandai Namco Research Inc.

Find here for a README in Japanese.

There is a long-standing interest in making diverse stylized motions for games and movies that pursue realistic and expressive character animation; however, creating new movements that include all the various styles of expression using existing methods is difficult. Due to this, Motion Style Transfer (MST) has been drawing attention recently, which aims to convert the motion in a clip with a given content into another motion in a different style, while keeping the same content. A motion is composed of a content and style, where content is the base of the motion and style comprises of the attributes such as mood and personality of the character tied to the motion.

The datasets contain a diverse range of contents such as daily activities, fighting, and dancing; with styles such as active, tired, and happy. These can be used as training data for MST models. The animation below shows examples of visualized motions.

Currently, two datasets are available in this repository and are both located under the dataset directory.

<img src="src/Bandai-Namco-Research-Motiondataset-1/movie_walk.gif" width="100%"> <img src="src/Bandai-Namco-Research-Motiondataset-2/movie_walk.gif" width="100%">

You can find a sample script for visualization (using Blender) including step-by-step instructions under the utils directory. Please see here for details.

Data Collection

Each dataset is based on the motion of three professional actors, collected at the motion capture studio of Bandai Namco. We applied post-processing such as noise removal, proportion alignment, and clipping; and saved in BVH format.

<img src='src/images/motion_capture_studio.png' width="50%"/>

Citation Information

The technical details regarding this dataset are described in the following paper.

If you use this dataset, please consider citing the paper:

Motion Capture Dataset for Practical Use of AI-based Motion Editing and Stylization

@misc{kobayashi2023motion,
      title={Motion Capture Dataset for Practical Use of AI-based Motion Editing and Stylization}, 
      author={Makito Kobayashi and Chen-Chieh Liao and Keito Inoue and Sentaro Yojima and Masafumi Takahashi},
      year={2023},
      eprint={2306.08861},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

License

The datasets and scripts are available in the following lincenses.

Related works

As mentioned previously, the field of MST has been drawing attention as of late. If you are interested in this topic, please refer to the following non-exhaustive list of research and datasets. The models introduced in these papers show a high-level of performance and the datasets related to MST provide a wide variety of motion data.

Papers

Datasets

© [2022] Bandai Namco Research Inc. All Rights Reserved