Qffusion: Controllable Portrait Video Editing via Quadrant-Grid Attention Learning

Maomao Li📌, Lijian Lin📌, Yunfei Liu, Ye Zhu, Yu Li

Project | Video

• We propose a novel dual-frame-guided framework for portrait video editing, which propagates fine-grained local modification from the start and end video frames.
• We propose a recursive inference strategy named Quadrant-grid Propagation (QGP), which can stably generate arbitrary-long videos.

GUAVA: Generalizable Upper Body 3D Gaussian Avatar

Dongbin Zhang, Yunfei Liu📩, Lijian Lin, Ye Zhu, Yang Li, Minghan Qin, Yu Li, Haoqian Wang📩

Project | Video

• ⚡️ Reconstructs 3D upper-body Gaussian avatars from single image in 0.1s
• ⏱️ Supports real-time expressive animation and novel view synthesis at 50FPS !

HRAvatar: High-Quality and Relightable Gaussian Head Avatar

Dongbin Zhang, Yunfei Liu, Lijian Lin, Ye Zhu, Kangjie Chen, Minghan Qin, Yu Li, Haoqian Wang

Project | Code

• We propose HRAvatar, a 3D Gaussian Splatting-based method that reconstructs high-fidelity, relightable 3D head avatars from monocular videos by jointly optimizing tracking, deformation, and appearance modeling.
• By leveraging learnable blendshapes, physically-based shading, and end-to-end optimization, HRAvatar significantly improves head quality and realism under novel lighting conditions.

TEASER: Token Enhanced Spatial Modeling for Expressions Reconstruction

Yunfei Liu, Lei Zhu, Lijian Lin, Ye Zhu, Ailing Zhang, Yu Li

Project | Code

• A novel approach that achieves more accurate facial expression reconstruction by predicting a hybrid representation of faces from a single image.
• A multi-scale facial appearance tokenizer and a token-guided neural renderer to generate high-fidelity facial images. The extracted token is interpretable and highly disentangled, enabling various downstream applications.

MODA: Mapping-Once Audio-driven Portrait Animation with Dual Attentions

Yunfei Liu, Lijian Lin, Fei Yu, Changyin Zhou, Yu Li

Project

• We propose a unified system for multi-person, diverse, and high-fidelity talking portrait video generation.
• Extensive evaluations demonstrate that the proposed system produces more natural and realistic video portraits compared to previous methods.

Accelerating the training of video super-resolution models

Lijian Lin, Xintao Wang, Zhongang Qi, Ying Shan

• Our method is capable of largely speeding up training (up to speedup in wall-clock training time) without performance drop for various VSR models.

Dual semantic fusion network for video object detection

Lijian Lin📌, Haosheng Chen📌, Honglun Zhang, Jun Liang, Yu Li, Ying Shan, Hanzi Wang

• We present a dual semantic fusion network, which performs a multi-granularity semantic fusion at both frame level and instance level in a unified framework and then generates enhanced features for video object detection.
• We introduce a geometric similarity measure into the proposed dual semantic fusion network along with the widely used appearance similarity measure to alleviate the information distortion caused by noise during the fusion process.