Page Not Found
Page not found. Your pixels are in another canvas. Read more
Sitemap
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Pages
From the First Token
Building language models from scratch, with worked examples and exercises. Read more
Page not in menu
This is a page not in th emain menu Read more
Posts
From Pixels to Systems: Practical Lessons from Generative Vision Research
Published:
Practical observations from work on image restoration, generative editing, efficient adaptation, assistive vision, and production machine learning systems. Read more
publications
High‐quality retinal vessel segmentation using generative adversarial network with a large receptive field
International Journal of Imaging Systems and Technology 30(3), 828-842, 2020
Probability-Based Channel Pruning for Depthwise Separable Convolutional Networks
Journal of Computer Science and Technology 37, 584-600, 2022
计算机科学技术学报 37(3)
基于概率的深度可分离卷积网络通道剪枝
计算机科学技术学报 37(3), 584-600, 2022
A Chinese journal record on probability-based channel pruning for depthwise separable convolutional networks.
计算机辅助设计与图形学学报 34(9)
基于感知去模糊的高效人脸图像修复算法
计算机辅助设计与图形学学报 34(9), 1420-1431, 2022
A Chinese journal article on efficient face image restoration with perceptual deblurring.
Journal of Computer-Aided Design & Computer Graphics 34(9)
Efficient Generative Face Completion with Perceptual Deblurring
Journal of Computer-Aided Design & Computer Graphics 34(9), 1420-1431, 2022
Efficient generative face completion with perceptual deblurring for higher-quality facial image restoration.
Generative image inpainting with enhanced gated convolution and Transformers
Displays 76, 102321, 2022
China Sciencepaper 17(11)
基于多尺度特征先验的图像拼接篡改检测
China Sciencepaper 17(11), 2022 · Second author
A Chinese journal article on image splicing tamper detection with multi-scale feature priors.
Restoring Vision in Hazy Weather with Hierarchical Contrastive Learning
Pattern Recognition Volume 145, January 2024, 109956
TEG: image theme recognition using text-embedding-guided few-shot adaptation
Journal of Electronic Imaging 33(1), 013028, 2024
FACEMUG: A Multimodal Generative and Fusion Framework for Local Facial Editing
IEEE Transactions on Visualization and Computer Graphics 31(9), 5130-5145, 2025 (online 2024) · First author
A multimodal generative framework for globally consistent local facial editing with fine-grained semantic control.
PhD thesis
Generative Models for Semantic Facial Image Editing: Multimodal Approaches
PhD thesis, Memorial University of Newfoundland, 2025
Doctoral thesis on generative models for semantic facial image editing and multimodal facial manipulation.
TextDoctor: Unified Document Image Inpainting via Patch Pyramid Diffusion Models
arXiv preprint, 2025 · First author · Submitted to Engineering Applications of Artificial Intelligence (EAAI)
A unified document image inpainting framework for restoring damaged and high-resolution document regions.
Degradation-Aware Frequency-Separated Transformer for Blind Super-Resolution
International Conference on Computational Visual Media, 2025
Visual style prompt learning using diffusion models for blind face restoration
Pattern Recognition 161, 111312, 2025 · First author
A diffusion-guided blind face restoration method that learns visual style prompts in pretrained generative latent space.
Real-time dual-eye collaborative eyeblink detection with contrastive learning
Pattern Recognition, 2025
Toward Biosignals-Free Autonomous Prosthetic Hand Control via Imitation Learning
IEEE Transactions on Neural Systems and Rehabilitation Engineering 33, 3544-3554, 2025
A biosignals-free autonomous prosthetic grasping framework that learns hand control through simulation-driven imitation learning.
EchoSR: Efficient Context Harnessing for Lightweight Image Super-Resolution
Information Fusion 135, 104471, 2026 · Corresponding author (sixth author)
An efficient context-harnessing framework for lightweight image super-resolution with local, multi-scale, and global modeling.
Simulation-Driven Imitation Learning for Biosignals-Free Shared-Autonomy Prosthetic Grasping
arXiv preprint arXiv:2606.07389, 2026
A simulation-driven imitation learning method for biosignals-free shared-autonomy prosthetic grasping.
UHDRes: Ultra-High-Definition Image Restoration via Dual-Domain Decoupled Spectral Modulation
IEEE Transactions on Circuits and Systems for Video Technology, 2026
A lightweight UHD restoration framework using dual-domain decoupled spectral modulation with only 400K parameters.
Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing
IEEE Transactions on Neural Networks and Learning Systems, 1-15, 2026 · First author
UltraDiffEdit extends off-the-shelf latent diffusion models to tuning-free real-image editing up to 8K resolution.
DocPure: Prompt-Free Unified Document Restoration via Degradation-Aware Structure-Guided Wavelet Modulation
IEEE Transactions on Circuits and Systems for Video Technology, 2026 · Co-first author
A prompt-free unified document restoration framework that adapts to diverse degradations through structure-aware routing and wavelet-domain feature interaction.