TalkBody4D Dataset

This dataset contains four multi-view image sequences used in our paper "TaoAvatar: Real-Time Lifelike Full-Body Talking Avatars for Augmented Reality via 3D Gaussian Splatting". They are captured with 59 well-calibrated RGB cameras in 20 fps, with a resolution of 3000×4000 and lengths ranging from 800 to 1000 frames. We use the data to evaluate our method for building animatable human body avatars.

We also provide the SMPL-X fitting in the dataset.

Agreement

Before we are able to offer you access to the database, please agree to the following terms of use.

After approval, you (the "Researcher") receive permission to use the TalkBody4D database (the "Database") at the Alibaba Group. In exchange for receiving permission to the Database, Researcher hereby agrees to the following terms and conditions:

Researcher shall use the Database only for non-commercial research and educational purposes. Any other use, in particular any use for commercial purposes, is prohibited. This includes, without limitation, incorporation in a commercial product, use in a commercial service, as training data for a commercial product, for commercial ergonomic analysis (e.g. product design, architectural design, etc.), or production of other artifacts for commercial purposes including, for example, web services, movies, television programs, mobile applications, or video games. The dataset may not be used for pornographic purposes or to generate pornographic material whether commercial or not. The Dataset may not be reproduced, modified and/or made available in any form to any third party without Alibaba Group’s prior written permission.
The Alibaba Group makes no representations or warranties regarding the Database, including but not limited to warranties of non-infringement or fitness for a particular purpose.
Researcher accepts full responsibility for his or her use of the Database and shall defend and indemnify the Alibaba Group, including their employees, Trustees, officers and agents, against any and all claims arising from Researcher's use of the Database.
Researcher may only provide research associates and colleagues with the data, after the receiving entity has also agreed to and signed these terms and conditions. Sharing the data otherwise is strictly prohibited.
You may however redistribute single images or derived videos from the Database as a part of an academic research paper, following the citing instructions that accompany the TalkBody4D Dataset. This includes videos, websites, or posts on social media platforms that accompany the paper’s publication.
As part of GDPR, Researcher must ensure that they can delete all person-specific data of a person upon request.
The Alibaba Group reserves the right to terminate Researcher’s access to the Database or parts thereof at any time. Reasons for this include, but are not limited to, GDPR requirements.
If Researcher is employed by a for-profit, commercial entity, Researcher's employer shall also be bound by these terms and conditions, and Researcher hereby represents that he or she is fully authorized to enter into this agreement on behalf of such employer.
The Chinese Law shall apply to all disputes under this agreement.

Download Instructions

The dataset is encrypted to prevent unauthorized access. Please fill out the request form : https://forms.gle/eC2aLRXZ8DAdKcis7

By requesting the link, you acknowledge that you have read the agreement, understand it, and agree to be bound by it. If you do not agree with these terms and conditions, you must not download and use the Dataset.

Process Steps:

Visit our HuggingFace repository and fill in the login details
Submit the corresponding request form.
We will verify both your account identity and form submission. Access will be enabled on HuggingFace upon approval.

Data Explanation

For each subject, we provide the multi-view images (images) as well as the matting (mattings). The calibration data is provided in calibration.json, and the SMPL-X fitting in (smplx-fitting).

Folder structure

|-- ROOT
    |-- Actor_ID #
        |-- raw_videos  #
            |-- 001.mp4
            |-- 002.mp4
            ....
        |-- mattings   # corresponding masks.
            |-- 001
                |-- 000000.png
                ....
            |-- 002
                |-- 000000.png
                ....
            ....
        |-- smplx_fitting # optimizes from multi-view images
            |-- 000000.json # json files
            |-- 000001.json
            ....
        
        |-- calibration.json   # extrinsics and intrinsics of all cameras

Convert raw_videos to images sequence:

ffmpeg -i raw_videos/${cam_ID}.mp4 -start_number 0 -pix_fmt rgb24 images/${cam_ID}/%06d.jpg

The camera extrinsics from calibration.json represent world-to-camera matrix in OpenCV coordinate system. The camera scale from calibration.json corresponds to the world scale, with the unit in meters.

To visualize smplx

Here we provide a code snip to show how to parse and visualize the data:

Download SMPLX_NEUTRAL.npz(about 228 MB) from https://github.com/vchoutas/smplx#downloading-the-model.
Put SMPLX_NEUTRAL.npz to the path of "assets/smplx_models"
run the "TalkBody4D_utils.py" script.

python TalkBody4D_utils.py \
    --images_folder {ROOT}/{Actor_ID}/images \  
    --params_folder ${ROOT}/{Actor_ID}/smplx_fitting \
    --calib_file ${ROOT}/{Actor_ID}/calibration.json \
    --model_path ../assets/smplx_models \
    --output_folder TalkBody4D_vis

Citation

If you find our work useful in your research, please consider citing:

Datasets:

PixelAI-Team
/

TalkBody4D

You need to agree to share your contact information to access this dataset