Zhedong Zheng (郑哲东)
Assistant Professor · University of Macau · AIGC-DL Lab
Better data, reliable AI. 更好的数据，更可靠的 AI
Hi! I am a tenure-track Assistant Professor at the University of Macau, hosting the AIGC-DL Lab. Previously a postdoctoral research fellow at NUS with Prof. Tat-Seng Chua and Prof. Angela Yao. I received my Ph.D. from the ReLER Lab, University of Technology Sydney (supervised by Prof. Yi Yang and Dr. Liang Zheng), and my Bachelor’s degree from Fudan University (2016). I have published 80+ papers at top-tier venues 
#3 Uncertainty Estimation
#5 Data-Centric AI
H-index 41
14k+ citations
Code ➞ GitHub · Publication ➞ SCI · SCOPUS · Google Scholar · CV
Open Positions
- 🎓 Ph.D. Students (Aug. 2027 intake, fully funded)
- 🔬 Postdocs (multiple funding lines, rolling basis / start anytime)
- 📚 Master Students (incoming Aug. 2026 intake; contact me early for full two-year research training)
Selected Talks
One invited talk in English, two public talks in Chinese — from person re-identification, to drone-view spatial intelligence, to multimedia UAVs for the low-altitude economy.
EN · INVITED TALK
Talk at 50:31
141 min
Aerial Spatial Intelligence: Perception, Understanding and Decision Making
ZH · INVITED TALK
31 min
Multimedia UAVs: Capturing the World from a New Perspective
ZH · INVITED TALK
58 min
From Person Re-Identification to Drone Localization
Research Philosophy
"Drones could serve as the third platform to deal with the geo-localization problem."
University-1652: A Multi-view Multi-source Benchmark for Drone-based Geo-localization · ACM MM 2020"We explicitly estimate the prediction uncertainty during training to rectify the pseudo label learning."
Rectifying Pseudo Label Learning via Uncertainty Estimation · IJCV 2021"Given the inherent cognitive uncertainty within reward models, we explicitly leverage the prediction variance as an uncertainty indicator."
Ctrl-U: Robust Conditional Image Generation via Uncertainty-aware Reward Modeling · ICLR 2025News
- PI, FDCT/0209/2025/AMJ, 2026–2029, MOP 1.66M
- ACM MM 2026: 2 papers - Dynamic Aerial Matching, Audio-Visual Seg.
- ECCV 2026 Workshop: AI City Challenge Track4
- ACL 2026: 2 papers - Multimodal Forgery Reasoning, Facial Forgery Report
- SIGIR 2026: Pretrain then Adapt
- ACM MM 2026 Workshop: Aerial-view Imaging [Call for Papers]
- TVCG 2026: 🎨 Look, Compare and Draw
- CVPR 2026: Manipulation Detection
- ICLR 2026: Efficient LLM
- Our work on multimedia aerial vision was showcased at the UM 45th Anniversary Open Day
Past News (2025-2020)
- NeurIPS 2025: 2 papers — WeatherPrompt, Motion Generation
- PI, FDCT/0043/2025/RIA1, 2025–2028, MOP 2.13M
- ACM MM 2025: 2 papers — Anomaly Detection, Neural Painting - BNI Track (Oral)
- ICCV 2025: 3 papers — Video2BEV, Anomaly Retrieval (Highlight), 3D Detection
- ICLR 2025: Conditional Generation
- ACM WWW 2025: Text-based Person Search
- ACM MM 2024: Domain Adaptation (Oral)
- ECCV 2024: 3 papers — Text-guided Geolocalization, Blind Image Decomposition, Unsupervised 3D Object Detection
- ACM MM 2025 Workshop: Aerial-view Imaging [Call for Papers]
- ACM MM Asia 2025 Brave New Idea Track: Call for Papers New
- ACM MM 2024 Workshops: Aerial-view Imaging, Multimodal Learning
- ICLR 2024: Compositional Retrieval [code]
- ACM ICMR 2024 Workshop: Multimedia Object Re-ID [Call for Papers]
- ACM MM 2023: 2 papers — Text-image Re-ID, Domain Adaptation
- CVPR 2023: 2 papers — Weather Removal, Video Moment Retrieval
- IJCV 2022: Adversarial Retrieval
- TNNLS 2022: 3D Human Re-ID
- TIP 2022: 2 papers — AdaBoost Adaptation, Drone Geolocalization
- CVPR 2022: NeRF
- ICME 2022: Special Session on Responsible Retrieval [Call for Papers]
- Ph.D. Thesis: Chancellor’s List
- Award: 2021 IEEE CAS Outstanding Young Author Award [link]
- IJCV 2021: Uncertainty [code]
- ACM MM 2020: Drone-view Dataset [PDF] [Dataset] [Video] [Slide]
Others
- Shanghai is my hometown, and it is a lovely place to have a sightseeing tour.
- I was a poster maker when I studied at Fudan University. You may check out my posters.
- I sometimes write Chinese blogs and share insights at Zhihu.
Do not press the red button!
