VIBE Lab
VIBE Lab · HUST
We develop generative intelligence systems that connect large-scale models with real-world perception and action.
We are recruiting PhD and master’s students and research assistants. Bachelor’s and high school students are also welcome to join projects in generative and embodied AI.
About
Prof. Tao Hu (胡涛) is a Full Professor at HUST whose research focuses on generative AI, including diffusion models, flow matching, and efficient generative architectures. He received his Ph.D. from the University of Amsterdam in 2023 (B.S., Wuhan University; M.S., Peking University), supervised by Prof. Cees Snoek at VISlab. He then conducted postdoctoral research with Prof. Björn Ommer at the CompVis Group, LMU Munich.
His work has appeared at CVPR, ICCV, ECCV, NeurIPS, ICLR, AAAI, and T-PAMI. Selected contributions include DepthFM (AAAI 2025 Oral), ZigMa (ECCV 2024), and Self-Guided Diffusion Models (CVPR 2023).
Latest News
| Nov 01, 2026 | Three new VIBE Lab preprints are now on arXiv: Search-Aware On-Policy Self-Distillation with Sibling Alignment for Generative Recommendation (Nov 1), GramLoop: Training-Free Gram-Gated Replay for Robust Dense Prediction (Aug 29), and Training-Free Hidden-State Refinement for Flow-Matching Image Generators (Aug 29). |
|---|---|
| Jul 27, 2026 | Selected as an Excellent Reviewer for ECCV 2026. |
| May 27, 2026 | Diversifying Similar Subjects for Text-to-image Synthesis with Self-Cross Diffusion Guidance is accepted by IJCV, congrats to Weimin, Jieke, and Meng. |
Research Areas
- World Model — Predictive representations for long-horizon video generation, planning, and embodied interaction
- Generative AI — Diffusion, flow matching, and efficient generative architectures
- Vision & Embodied AI — Large vision models, multimodal learning, robot learning, and sim-to-real embodied intelligence
What We Offer
- Supportive and collaborative research environment
- Strong mentorship and international collaboration network
- Access to substantial computing resources
- Codex API reimbursement
- Opportunities to publish at leading venues such as CVPR, ICCV, NeurIPS, and ICLR
Group Members
Chengsong Qu
Canyu Shen