扫码或长按图片保存 · Scan the QR in RedNote, or long-press to save
Ensure that artificial general intelligence benefits all of humanity.
— OpenAI Mission · the north star I work toward
.png){kind=logo}
Reasoning Denoiser: Denoising Reasoning Traces for Hallucination Detection in Large Reasoning Models
Last But Not Least: Boundary Attention CalibratiON for Multimodal KV Cache Compression
HDRAgent: An Agentic Framework for Multi-Exposure HDR Imaging
Streaming Communication in Multi-Agent Reasoning
LITMUS: Benchmarking Behavioral Jailbreaks of LLM Agents in Real OS Environments
AHPA: Adaptive Hierarchical Prior Alignment for Diffusion Transformers
From Thinker to Society: Security in Hierarchical Autonomy Evolution of AI Agents
Low-Light Video Enhancement with An Effective Spatial-Temporal Decomposition Paradigm
MiroThinker-1.7 & H1: Towards Heavy-Duty Research Agents via Verification
Multi-Agent Tool-Integrated Policy Optimization with Process Reward
Enhancing Diffusion-based Restoration Models via Difficulty-Adaptive Reinforcement Learning with IQA Reward
Boosting Fidelity for Pre-Trained-Diffusion-Based Low-Light Image Enhancement via Condition Refinement
MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling
Jailbreaking Commercial Black-Box LLMs with Explicitly Harmful Prompts
Generative Distribution Distillation
Generative Visual Commonsense Answering and Explaining with Generative Scene Graph Constructing
From Events to Enhancement: A Survey on Event-Based Imaging Technologies
Iter-AHMCL: Alleviate Hallucination for Large Language Model via Iterative Model-level Contrastive Learning
Self-supervised Learning for Enhancing Geometrical Modeling in 3D-Aware Generative Adversarial Network
Clarity ChatGPT: An Interactive and Adaptive Processing System for Image Restoration and Enhancement
General Adversarial Defense Against Black-box Attacks via Pixel Level and Feature Level Distribution Alignments
SpeedVFI: One-step Diffusion for Efficient Video Frame Interpolation
GDRO: Group-level Reward Post-training Suitable for Diffusion Models
Learning Latent Proxies for Controllable Single-Image Relighting
LoViC: Efficient Long Video Generation with Context Compression
Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance
DiffDoctor: Diagnosing Image Diffusion Models Before Treating
LucidDreamer: Towards High-Fidelity Text-to-3D Generation via Interval Score Matching
Conditional Temporal Variational AutoEncoder for Action Video Prediction
Hierarchical Image Generation via Transformer-Based Sequential Patch Selection
Text-Guided Human Image Manipulation via Image-Text Shared Space
Multi-Agent Tool-Integrated Policy Optimization with Process Reward
MiroThinker-1.7 & H1: Towards Heavy-Duty Research Agents via Verification
LPO: Towards Accurate GUI Agent Interaction via Location Preference Optimization
ACIArena: Toward Unified Evaluation for Agent Cascading Injection
PerMemSafe: Benchmarking Implicit Personalized Safety of Long Horizon Self-Evolving Agents
Fair in Mind, Fair in Action? A Synchronous Benchmark for Understanding and Generation in UMLLMs
Robust-R1: Degradation-Aware Reasoning for Robust Visual Understanding
MiCo: Multi-image Contrast for Reinforcement Visual Reasoning
LARM: Large Auto-Regressive Model for Long-Horizon Embodied Intelligence
Hawk: Learning to Understand Open-World Video Anomalies
Learnable Feature Patches and Vectors for Boosting Low-light Image Enhancement without External Knowledge
Boosting Image Restoration via Priors from Pre-trained Models
Geometric-Aware Low-Light Image and Video Enhancement via Depth Guidance
SNR-Aware Low-light Image Enhancement
Low‑light Image Enhancement via Structure Modeling and Guidance
Low-Light Video Enhancement via Spatial-Temporal Consistent Decomposition
Deep Parametric 3D Filters for Joint Video Denoising and Illumination Enhancement in Video Super Resolution
Seeing Dynamic Scene in the Dark: A High-Quality Video Dataset with Mechatronic Alignment
Depth Anything V2
Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data
TriVol: Point Cloud Rendering Via Triple Volumes
Point2Pix: Photo‑Realistic Point Cloud Rendering via Neural Radiance Fields
Universal Adaptive Data Augmentation
MTFormer: Multi-Task Learning via Transformer and Cross-Task Reasoning
Dynamic divide-and-conquer adversarial training for robust semantic segmentation
HoloFair: Unified T2I Fairness Evaluation and Fair-GRPO Debiasing
SpeedVFI: One-step Diffusion for Efficient Video Frame Interpolation
Remember-R1: Mitigating Long-Context Visual Forgetting through Reinforcement Learning
Deconstructing Supervision: Rethinking Learning Signals in Class Incremental Medical Image Segmentation
Hybrid Event–Frame Sensors: Modeling, Calibration, and Simulation
ACIArena: Toward Unified Evaluation for Agent Cascading Injection
PerMemSafe: Benchmarking Implicit Personalized Safety of Long Horizon Self-Evolving Agents
LPO: Towards Accurate GUI Agent Interaction via Location Preference Optimization
Distinguish Any Fake Videos: Unleashing the Power of Large-scale Data and Motion Features
IAMAgent: Towards An Interactive and Adaptive Multi-Agent System for Image Restoration
GDRO: Group-level Reward Post-training Suitable for Diffusion Models
Learning Latent Proxies for Controllable Single-Image Relighting
LoViC: Efficient Long Video Generation with Context Compression
Fair in Mind, Fair in Action? A Synchronous Benchmark for Understanding and Generation in UMLLMs
FedGraph: Defending Federated Large Language Model Fine-Tuning Against Backdoor Attacks via Graph-Based Aggregation
GADT: Enhancing Transferable Adversarial Attacks through Gradient-guided Adversarial Data Transformation
FC-VFI: Faithful and Consistent Video Frame Interpolation for High-FPS Slow Motion Video Generation
Boosting HDR Image Reconstruction via Semantic Knowledge Transfer
Class Incremental Medical Image Segmentation via Prototype-Guided Calibration and Dual-Aligned Distillation
Robust-R1: Degradation-Aware Reasoning for Robust Visual Understanding
Restoration-Oriented Video Frame Interpolation with Region-Distinguishable Priors from SAM
SEE: See Everything Every Time--Adaptive Brightness Adjustment for Broad Light Range Images via Events
Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance
MiCo: Multi-image Contrast for Reinforcement Visual Reasoning
DiffCamera: Arbitrary Refocusing on Images
PRIME: Prototype-Driven Class Incremental Learning for Medical Image Segmentation
PVDD: A Practical Benchmark Dataset and Network for Video Denoising
Geometric-Aware Low-Light Image and Video Enhancement via Depth Guidance
Boosting Diffusion-Based Text Image Super-Resolution Model Towards Generalized Real-World Scenarios
Diffusion Noise Feature: Accurate and Fast Generated Image Detection
CG-FedLLM: How to Compress Gradients in Federated Fune-tuning for Large Language Models
Learnable Feature Patches and Vectors for Boosting Low-light Image Enhancement without External Knowledge
Co-Painter: Fine-Grained Controllable Image Stylization via Implicit Decoupling and Adaptive Injection
DiffDoctor: Diagnosing Image Diffusion Models Before Treating
Adversarial Attacks of Vision Tasks in the Past 10 Years: A Survey
Towards Unified 3D Object Detection via Algorithm and Data Unification
FashionComposer: Compositional Fashion Image Generations
CFSynthesis: Controllable and free-view 3d human video synthesis
DiMSOD: A Diffusion-Based Framework for Multi-Modal Salient Object Detection
DR-Encoder: Encode Low-rank Gradients with Random Prior for Large Language Models Differentially Privately
Towards Better Adversarial Purification via Adversarial Denoising Diffusion Training
Diving Deep into Regions: Exploiting Regional Information Transformer for Single Image Deraining
Low-Light Video Enhancement via Spatial-Temporal Consistent Decomposition
LARM: Large Auto-Regressive Model for Long-Horizon Embodied Intelligence
Sagiri: Low Dynamic Range Image Enhancement with Generative Diffusion Prior
Particle Rendering: Implicitly Aggregating Incident and Outgoing Light Fields for Novel View Synthesis
HELPD: Mitigating Hallucination of LVLMs by Hierarchical Feedback Learning with Vision-enhanced Penalty Decoding
Parametric Linear Blend Skinning Model for Multiple-Shape 3D Garments
Hawk: Learning to Understand Open-World Video Anomalies
Depth Anything V2
Unveiling Advanced Frequency Disentanglement Paradigm for Low-Light Image Enhancement
An Incremental Unified Framework for Small Defect Inspection
Refine, Discriminate and Align: Stealing Encoders via Sample-Wise Prototypes and Multi-Relational Extraction
Towards Efficient Large-Scale Language-3D Representation Learning
Generative Active Learning for Long-tailed Instance Segmentation
Depth Anything: Unleashing the Power of Large-Scale Unlabeled Data
MalGNE: Enhancing the Performance and Efficiency of CFG-based Malware Detector by Graph Node Embedding in Low Dimension Space[Paper] Code
LucidDreamer: Towards High-Fidelity Text-to-3D Generation via Interval Score Matching
UniMODE: Universal Monocular 3D Object Detection
Boosting Image Restoration via Priors from Pre-trained Models
Learning to Remove Wrinkled Transparent Film with Polarized Prior
S2WAT: Image Style Transfer via Hierarchical Vision Transformer using Strips Window Attention
Densely Annotated Synthetic Images Make Stronger Semantic Segmentation Models
CorresNeRF: Image Correspondence Priors for Neural Radiance Fields
Photo-Realistic Out-of-domain GAN inversion via Invertibility Decomposition
Lighting up NeRF via Unsupervised Decomposition and Enhancement
High Dynamic Range Image Reconstruction via Deep Explicit Polynomial Curve Estimation
Low‑light Image Enhancement via Structure Modeling and Guidance
TriVol: Point Cloud Rendering Via Triple Volumes
Point2Pix: Photo‑Realistic Point Cloud Rendering via Neural Radiance Fields
Deep Parametric 3D Filters for Joint Video Denoising and Illumination Enhancement in Video Super Resolution
Conditional Temporal Variational AutoEncoder for Action Video Prediction
Universal Adaptive Data Augmentation
MTFormer: Multi-Task Learning via Transformer and Cross-Task Reasoning
DecoupleNet: Decoupled Network for Domain Adaptive Semantic Segmentation
SNR-Aware Low-light Image Enhancement
Hierarchical Image Generation via Transformer-Based Sequential Patch Selection
Dynamic divide-and-conquer adversarial training for robust semantic segmentation
Seeing Dynamic Scene in the Dark: A High-Quality Video Dataset with Mechatronic Alignment
Text-Guided Human Image Manipulation via Image-Text Shared Space
Adversarial Captchas (Applied and deployed on Alibaba's e-commerce platform) 
Self-Supervised 3D Mesh Reconstruction from Single Images
Semantic-Aware Video Color Style Transfer based on Temporal Consistent Sparse Patch Constraint[Paper] Code
Reference-based Video Colorization with Multi-scale Semantic Fusion and Temporal Augmentation[Paper] Code
Domain Adaptive Image-to-image Translation
View Independent Generative Adversarial Network for Novel View Synthesis[Paper] Code
Homomorphic Latent Space Interpolation for Unpaired Image-to-image Translation
Ranking Users in Social Networks with Motif-based PageRank
Ranking Users in Social Networks with Higher-Order Struct
No‑reference stereoscopic image quality assessment based on saliency‑guided binocular feature consolidation
Research is what I do; these are what keep me whole. 研究之外，我更向往生活本身。
Long distances teach patience better than any paper deadline — one steady step at a time.
长跑教会我的耐心，比任何 deadline 都多。
Every summit is a reminder that the view is worth the climb.
山顶的风景，总值得那段上坡路。
Borrowing other lives and other centuries, a few pages at a time.
读书，是借别人的一生，换自己的开阔。
Sitting with the questions that have no benchmark and no leaderboard.
有些问题没有 benchmark，只能慢慢想、慢慢悟。
「路虽远，行则将至；事虽难，做则必成。」
Scholar
LinkedIn
GitHub
