I-HSIANG (AARON) CHEN

Bridging vision,
generation, and
the real world.

I explore how machines perceive, restore, and understand the visual world. My research connects generative models, robust visual learning, and multimodal understanding, with a focus on systems that work beyond clean, familiar conditions.

Explore my research View CV

Computer visionGenerative AIMultimodal learning

Portrait of I-Hsiang (Aaron) Chen — I-Hsiang (Aaron) ChenPerception. Generation. Understanding.

RESEARCH ACROSS ACADEMIA & INDUSTRY

National Taiwan UniversityUniversity of WashingtonSamsung ResearchMediaTekASML

EXPERIENCE

A professional journey.

My research background spans National Taiwan University and the University of Washington, alongside industry research experience at Samsung Research, MediaTek, and ASML.

Scroll horizontally to explore the full journey.

SELECTED RESEARCH

From questions to contributions.

All publications

Image restoration for human perception and downstream vision tasks.

CVPR 2025✳ Highlight

UniRestore

Better to our eyes. Better for the task.

Image Generation and EditingMultimodal LearningVisual Understanding

Paper Project Watch video

A visual comparison of baseline and PDAF semantic segmentation predictions.

ICCV 2025

PDAF

Understanding scenes beyond familiar conditions.

Domain GeneralizationVisual Understanding

Paper Project Watch video

Crowd counting and localization with point detections in a real scene.

ECCV 2024

APGCC

Every point deserves a better match.

Visual Understanding

Paper Project Watch video

Iterative quality assessment and restoration: recovering detail from a degraded image.

CVPR 2026

Restore. Assess. Repeat.

Restoration that knows when to look again.

Image Generation and EditingMultimodal Learning

Paper Project Watch video

Retrieving the relevant visual document despite degraded image quality. — Retrieve

Comparing generated answers to a question about a degraded retrieved document. — Generation

CVPR 2026

RobustVisRAG

Reliable answers start with resilient perception.

Multimodal LearningVisual Understanding

Paper Project Watch video

Swipe or scroll to explore all five projects. Arrow buttons move one project at a time.

LATEST UPDATES

What's new.

Research updates, new experiences,
and milestones along the way.

Apr 2026PublicationRestore, Assess, Repeat has been accepted to CVPR 2026.Apr 2026PublicationRobustVisRAG has been accepted to CVPR 2026.Nov 2025RecognitionRecognized with the NTU Outstanding Young Award.Jul 2025PublicationExploring probabilistic modeling for semantic segmentation at ICCV 2025.Jun 2025ExperienceJoined Samsung Research UK for a research internship.

Earlier updates

Apr 2025PublicationUniRestore selected as a CVPR 2025 Highlight.Jul 2024PublicationAuxiliary Point Guidance for crowd counting at ECCV 2024.May 2024ExperienceBegan visiting research at the University of Washington.Dec 2023RecognitionReceived the CTCI Research Award.

RECOGNITION

Milestones
along the way.

NTU Outstanding Young Award

National Taiwan University

Explore awards 2023

CTCI Research Award

CTCI Foundation