👋 I am currently an Associate Professor at the Institute of Artificial Intelligence, Xiamen University, affiliated with the MAC Lab. My research focuses on efficient multimodal models, with the ultimate goal of achieving physical intelligence through highly efficient multimodal perception, i.e., multi-modal physical intelligence with efficiency.
🧑🔬 I have extensive experience in both academia and industry. I previously worked as a Research Fellow at NUS xML Lab, collaborating with Professor Xinchao Wang. I also served as an engineer and postdoctoral researcher at Contemporary Amperex Technology Co. Limited (CATL), where I worked with Dr. Guannan Jiang under the guidance of Professor Jun Ni. I earned my Ph.D. from Xiamen University, supervised by Professor Rongrong Ji and Professor Liujuan Cao.
📖 I am constantly seeking self-motivated Master’s and Ph.D. students. If you are interested in joining my group, please read my Research Intro first to assess whether we are a good fit for each other.
🔥 News
- 2026.03: 🎉 I joined Xiamen University as an Associate Professor.
- 2026.02: 🚀 One paper accepted with CVPR’26.
📝 Selected Publications
PE3R: Perception-Efficient 3D Reconstruction
Jie Hu, Shizun Wang, Xinchao Wang
PE3R reconstructs 3D scenes using only 2D images and enables semantic understanding through language.
Universal Image Segmentation With Efficiency
Jie Hu, Liujuan Cao, Xiaofeng Jin, et al.
UISE is an unified segmentation framework that achieves competitive speed and accuracy across multiple tasks through dynamic convolutions, eliminating the need for specialized pipelines.
ISTR: Mask-Embedding-Based Instance Segmentation Transformer
Jie Hu, Yao Lu, Shengchuan Zhang, Liujuan Cao
This paper proposes a transformer-based instance segmentation framework, which encodes masks into embeddings to regress them.
-
Pseudo-label Alignment for Semi-supervised Instance Segmentation. Jie Hu, Chen Chen, Liujuan Cao, et al. ICCV’23 [code]
-
You Only Segment Once: Towards Real-Time Panoptic Segmentation. Jie Hu, Linyan Huang, Tianhe Ren, et al. CVPR’23 [code]
-
DistilPose: Tokenized Pose Regression with Heatmap Distillation. Suhang Ye, Yingyi Zhang, Jie Hu, et al. CVPR’23 [code]
-
Architecture Disentanglement for Deep Neural Networks. Jie Hu, Liujuan Cao, Qixiang Ye, et al. ICCV’21, oral [code]
-
Image-to-image Translation via Hierarchical Style Disentanglement. Xinyang Li, Shengchuan Zhang, Jie Hu, et al. CVPR’21, oral [code]
-
Towards Visual Feature Translation. Jie Hu, Rongrong Ji, Hong Liu, et al. CVPR’19, oral [code]
✨ Honors and Awards
- 2025.01 Fujian Provincial Excellent Doctoral Dissertation.
- 2015.11 Bronze Medal, ACM-ICPC Programming Contest Asia Regional - Hefei Site.
