About Me
I am currently a Researcher and Engineer in Zoom, based in Singapore. Before joining Zoom, I was a Postdoctoral Fellow in the Hong Kong Polytechnic University, supervised by Prof. Kin-Man Lam. Before PhD graduation, I was interned in Microsoft Research Asia.
Research Interests
- AIGC (post-training): image/video generation, world model, efficient generative models, and preference optimization.
- Low-level vision processing: image/video super-resolution, image/video denoising, and high dynamic range imaging.
- 3D vision: generative-based 3D reconstruction
I am actively seeking opportunities for research collaboration. Please feel free to reach out via email at jun.xiao@connect.polyu.hk.
Work Experience
| 2024 – Present | Research Engineer, Zoom Video Communications, Singapore Project 1: Logo-driven image generation — subject-consistency score 0.855; generation within 2s at 1280×720. Project 2: Text-driven style transfer for human images — human-ID consistency 0.85; style consistency 0.63; generation within 1s at 512×512. |
| 2022 – 2024 | Postdoctoral Fellow, The Hong Kong Polytechnic University, Hong Kong SAR Research project: Led the project “efficient old movie 4K photorealistic restoration” — designed an efficient video restoration model for low-resolution old movies without ground truth. Results: 4× faster processing; ~85% acceptance rate; delivered a cinema-released film (Nomad). |
| 2021 – 2022 | Research Intern, Microsoft Research Asia, Shanghai Research project: Involved in the project “online video restoration and enhancement system” — designed an efficient, lightweight super-resolution model for online streaming videos. Results: 110 FPS for 720p videos on a 2080Ti GPU; only 1.75M model parameters; published in IEEE Transactions on Multimedia. Award: MSRA Stars of Tomorrow (Award for Excellent Intern). |
Download my CV .
Education
Ph.D., The Hong Kong Polytechnic University (PolyU), 2018.09 -- 2022.10
M.S.c (with distinction), The Hong Kong Polytechnic University (PolyU), 2016.09 -- 2018.03
BSc, Guangdong University of Technology, 2012.09 -- 2016.06
Publications
(2026).
Dynamic Mutual Learning for Object Detection in Aerial Imagery.
In IEEE TGRS.
(2026).
Exploring Basic Expression Representation for Compound Facial Expression Recognition.
In IEEE TMM.
(2025).
Geometric Distortion Guided Transformer for Omnidirectional Image Super-resolution.
In IEEE TCSVT.
(2025).
See in Detail: Enhancing Sparse-view 3D Gaussian Splatting with Local Depth and Semantic Regularization.
In ICASSP.
(2025).
Photometric Regularization for 3D Gaussian Splatting in Multi-view Surface Projection.
In IEEE J-STSP.
(2024).
Point Cloud Densification for 3D Gaussian Splatting from Sparse Input Views.
In ACM MM.
(2024).
Bridging Text and Image for Artist Style Transfer via Contrastive Learning.
In ECCV-W.
(2024).
Frequency-Aware Guidance for Blind Image Restoration via Diffusion Models.
In ECCV-W.
(2024).
Towards Multi-View Consistent Style Transfer with One-Step Diffusion via Vision Conditioning.
In ECCV-W.
(2024).
Towards Progressive Multi-Frequency Representation for Image Warping.
In CVPR.
(2024).
Deep Progressive Feature Aggregation Network for Multi-frame High Dynamic Range Imaging.
In Neurocomputing.
(2024).
Deep Multi-scale Feature Mixture Model for Image Super-resolution with Multiple-Focal-length Degradation.
In Signal Processing: Image Communication.
(2023).
Improving Robustness of Single Image Super-Resolution Models with Monte Carlo Method.
In ICIP.
(2023).
Boosting Object Detectors via Strong-Classification Weak-Localization Pretraining in Remote Sensing Imagery.
In IEEE Transactions on Instrumentation and Measurement (TIM).
(2023).
Online Video Super-Resolution with Convolutional Kernel Bypass Grafts.
In IEEE Transaction on Multimedia (TMM).
(2023).
Efficient Feature Fusion for Learning-based Photometric Stereo.
In ICASSP.
(2021).
Self-feature Learning: An Efficient Deep Lightweight Network for Image Super-resolution.
In ACM-MM.
(2021).
Progressive and Selective Fusion Network for High Dynamic Range Imaging.
In ACM-MM.
(2021).
Balanced distortion and perception in single-image super-resolution based on optimal transport in wavelet domain.
In Neurocomputing.
(2021).
Invertible image decolorization.
In IEEE Transactions on Image Processing (TIP).
(2021).
Feature redundancy mining: Deep light-weight image super-resolution model.
In ICASSP.
(2021).
Bayesian sparse hierarchical model for image denoising.
In Signal Processing:Image Communication.
(2020).
Deep multi-task learning for facial expression recognition and synthesis based on selective feature sharing.
In ICPR.
(2020).
Progressive Motion Representation Distillation With Two-Branch Networks for Egocentric Activity Recognition.
In Signal Processing Letter.
Services
Journal Reviwer
- IEEE Transactions on Image Processing
- IEEE Transactions on Multimedia
- IEEE Transactions on Circuits and Systems for Video Technology.
- Knowledge-based Systems
- Information Fusion
- Pattern Recognition
Conference Reviwer
- Neural Information Processing Systems (NeurlPS)
- European Conference on Computer Vision (ECCV)
- ACM Conference on Multimedia (ACM-MM)
- Asia Conference on Computer Vision (ACCV)
- Winter Conference on Applications of Computer Vision (WACV)
- IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
- IEEE International Conference on Image Processing (ICIP)
- IEEE International Conference on Visual Communications and Image Processing (VCIP)