Kyeonghyun Lee

Kyeonghyun Lee

Auto-generated profile, based on publicly available information

Kyeonghyun Lee

Department of Electrical and Computer Engineering, Seoul National University, Seoul, Republic of Korea

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
20 January 2025 in IEEE Access
Verified
Journal Article
Open access
JLJoonhee Lee
Joonhee Lee
DPDongwon Park
Dongwon Park
SCSe Young Chun
Se Young Chun
Image quality assessment (IQA) is a fundamental task in computer vision with the goal of accurately predicting the mean opinion score of humans for assessing the quality of images. While recent advances in deep neural networks (DNNs) have sparked much research on IQA, with the hope for IQA to mimic humans effectively, there has been a lack of systematic and analytical research on understanding what factors humans prioritize during IQA. This paper aims to identify human priorities in image evaluation by leveraging the DNN models for diverse computer vision tasks and proposes simple, but effective IQA metrics through our comprehensive analyses on those models. Then, these analyses led us to propose a novel vision-ensemble IQA (VE-IQA) method that demonstrated superior performance as compared to prior arts in IQA on popular IQA benchmarks such as LIVE, CSIQ, and TID2013.

9 June 2025 in arXiv
Verified
Preprint
Open access
PDF file
JBJunseo Bang
Junseo Bang
JLJoonhee Lee
Joonhee Lee
KLKyeonghyun Lee
Kyeonghyun Lee
HLHaechang Lee
Haechang Lee
DKDong Un Kang
Dong Un Kang
SCSe Young Chun
Se Young Chun
Arbitrary-scale image super-resolution aims to upsample images to any desired resolution, offering greater flexibility than traditional fixed-scale super-resolution. Recent approaches in this domain utilize regression-based or generative models, but many of them are a single-stage upsampling process, which may be challenging to learn across a wide, continuous distribution of scaling factors. Progressive upsampling strategies have shown promise in mitigating this issue, yet their integration with diffusion models for flexible upscaling remains underexplored. Here, we present CasArbi, a novel self-cascaded diffusion framework for arbitrary-scale image super-resolution. CasArbi meets the varying scaling demands by breaking them down into smaller sequential factors and progressively enhancing the image resolution at each step with seamless transitions for arbitrary scales. Our novel coordinate-guided residual diffusion model allows for the learning of continuous image representations while enabling efficient diffusion sampling. Extensive experiments demonstrate that our CasArbi outperforms prior arts in both perceptual and distortion performance metrics across diverse arbitrary-scale super-resolution benchmarks.