no code implementations • 17 Dec 2023 • Congyi Wang
This vector then serves as a context prompt for the residual GPT, which generates the final residual discrete indices.
no code implementations • 29 Jul 2023 • Zuyan Liu, Gaojie Lin, Congyi Wang, Min Zheng, Feida Zhu
Our approach involves a unified and multi-granularity strategy that includes a pseudo keypoint alignment module in the teacher-student framework for learning pose-aware semantic class tokens.
no code implementations • CVPR 2023 • Congyi Wang, Feida Zhu, Shilei Wen
Existing methods proposed for hand reconstruction tasks usually parameterize a generic 3D hand model or predict hand mesh positions directly.
no code implementations • 22 Jun 2021 • Liguo Jiang, Miaopeng Li, Jianjie Zhang, Congyi Wang, Juntao Ye, Xinguo Liu, Jinxiang Chai
We show the system advances the frontier of 3D human body and pose reconstruction from single images by quantitative evaluations and comparisons with state-of-the-art methods.
no code implementations • 26 Mar 2021 • Congyi Wang, Yu Chen, Bin Wang, Yi Shi
GAN-based neural vocoders, such as Parallel WaveGAN and MelGAN have attracted great interest due to their lightweight and parallel structures, enabling them to generate high fidelity waveform in a real-time manner.
no code implementations • 1 Feb 2021 • Yi Shi, Congyi Wang, Yu Chen, Bin Wang
In this paper, we propose a novel semi-supervised learning (SSL) framework for Mandarin Chinese polyphone disambiguation that can potentially leverage unlimited unlabeled text data.
no code implementations • 2 Feb 2020 • Wenlin Zhuang, Congyi Wang, Siyu Xia, Jinxiang Chai, Yangang Wang
Synthesize human motions from music, i. e., music to dance, is appealing and attracts lots of research interests in recent years.
Ranked #5 on Motion Synthesis on AIST++