Teng Hu

Hi! I am a Ph.D. student from Shanghai Jiao Tong University (SJTU), under the supervision of Prof. Ran Yi in Digital Media & Computer Vision Laboratory (DMCV). My research interests mainly lie in Computer Vision and Generative Models, supported by National Natural Science Foundation of China for Young Ph.D. students (国家自然科学基金青年学生基础研究项目(博士生)) and CIE-Tencent Doctoral Research Incentive Project (首届中国电子学会—腾讯博士生科研激励计划(混元大模型专项)).

Email  /  Google Scholar  /  Github  / 

Internship. We am looking for self-motivated students to join our research group! If you are interested in our research, feel free to contact me!

[2025.11.8]: Our papers UltraGen is accepted by AAAI 2026!

[2025.9.30]: Our papers AttnPainter is accepted by TVCG 2025!

[2025.9.18]: Our papers PolyVivid, UltraVideo are accepted by NeurIPS 2025!

[2025.7.26]: Our papers EmoV2 is accepted by TPAMI 2025!

[2025.2.27]: Our papers IAR, ATA, DualAnoDiff, are accepted by CVPR 2025!

[2025.1.22]: Our paper SaRA: High-Efficient Diffusion Model Fine-tuning with Progressive Sparse Low-Rank Adaptation is accepted by ICLR 2025!

[2024.7.24]: Our paper FEditNet++: Few-Shot Editing of Latent Semantics in GAN Spaces with Correlated Attribute Disentanglement is accepted by TPAMI 2024!

[2024.7.16]: Three papers are accepted by ACM Multimedia 2024!

[2024.2.27]: Our paper SuperSVG: Superpixel-based Scalable Vector Graphics Synthesis is accepted by CVPR 2024!

[2023.12.12]: Our paper SAMVG: A Multi-stage Image Vectorization Model with the Segment-Anything Model is accepted by ICASSP 2024!

[2023.12.09]: Our paper AnomalyDiffusion: Few-Shot Anomaly Image Generation with Diffusion Model is accepted by AAAI 2024!

[2023.12.05]: 我们的论文 《基于薄板样条插值的弯曲笔触神经绘画与风格化方法》 被中国科学：信息科学接收! Our paper Curve-Stroke-Based Neural Painting and Stylization with Thin Plate Spline Interpolation is accepted by SCIENTIA SINICA Informationis!

[2023.07.26]: Our paper Stroke-based Neural Painting and Stylization with Dynamically Predicted Painting Region is accepted by ACM MM 2023!

[2023.07.14]: Our paper Phasic Content Fusing Diffusion Model with Directional Distribution Consistency for Few-Shot Model Adaption is accepted by ICCV 2023!

We propose MotionMaster, a novel training-free video motion transfer model, which disentangles camera motions and object motions in source videos, and transfers the extracted camera motions to new videos. We introduce a one-shot and few-shot camera motion disentanglement method, and design a camera motion combination method, enabling our model a more controllable and flexible camera control.

We propose AesStyler, a novel Aesthetic Guided Universal Style Transfer method, which utilizes pre-trained aesthetiic assessment model, a novel Universal Aesthetic Codebook and a novel Universal and Specific Aesthetic-Guided Attention (USAesA) module. Extensive experiments and user-studies have demonstrated that our approach generates aesthetically more harmonious and pleasing results than the state-of-the-art methods.

We propose SuperSVG, a superpixel-based vectorization model that achieves fast and high-precision image vectorization. we decompose the input image into superpixels to help the model focus on areas with similar colors and textures. Then, we propose a two-stage self-training framework, where a coarse-stage model is employed to reconstruct the main structure and a refinement-stage model is used for enriching the details.

We propose SAMVG, a multi-stage model to vectorize raster images into SVG (Scalable Vector Graphics). Through a series of extensive experiments, we demonstrate that SAMVG can produce high quality SVGs in any domain while requiring less computation time and complexity compared to previous state-of-the-art methods.

we propose AnomalyDiffusion, a novel diffusion-based few-shot anomaly generation model, which utilizes the strong prior information of latent diffusion model learned from large-scale dataset to enhance the generation authenticity under few-shot training data.

We propose a new curved brushstroke parameter model based on thin-plate spline interpolation. By curving and affine-transforming real brushstroke templates in succession, we can generate more realistic and varied brushstroke images. Furthermore, we propose a hierarchical brushstroke optimization method that decomposes the entire image into multiple brushstrokes, from large to small, effectively improving the model’s painting ability for both the overall structure and local details of the image

We propose Compositional Neural Painter, a novel stroke-based rendering framework which dynamically predicts the next painting region based on the current canvas, instead of dividing the image plane uniformly into painting regions. Extensive experiments show our model outperforms the existing models in stroke-based neural painting.

We propose a novel phasic content fusing few-shot diffusion model with directional distribution consistency loss, which targets different learning objectives at distinct training stages of the diffusion model. Theoretical analysis, and experiments demonstrate the superiority of our approach in few-shot generative model adaption tasks.