6 papers
Position: Evaluation of Visual Processing Should Be Human-Centered, Not Metric-Centered
Jinfan Hu, Fanghua Yu, Zhiyuan You +5
This position paper argues that the evaluation of modern visual processing systems should no longer be driven primarily by single-metric image quality assessment benchmarks, partic…
SimpleGVR: A Simple Baseline for Latent-Cascaded Video Super-Resolution
Liangbin Xie, Yu Li, Shian Du +7
Latent diffusion models have emerged as a leading paradigm for efficient video generation. However, as user expectations shift toward higher-resolution outputs, relying solely on l…
Harnessing Diffusion-Yielded Score Priors for Image Restoration
Xinqi Lin, Fanghua Yu, Jinfan Hu +5
Deep image restoration models aim to learn a mapping from degraded image space to natural image space. However, they face several critical challenges: removing degradation, generat…
Interpreting Low-level Vision Models with Causal Effect Maps
Jinfan Hu, Jinjin Gu, Shiyao Yu +5
Deep neural networks have significantly improved the performance of low-level vision tasks but also increased the difficulty of interpretability. A deep understanding of deep model…
UniCon: Unidirectional Information Flow for Effective Control of Large-Scale Diffusion Models
Fanghua Yu, Jinjin Gu, Jinfan Hu +2
We introduce UniCon, a novel architecture designed to enhance control and efficiency in training adapters for large-scale diffusion models. Unlike existing methods that rely on bid…
Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration In the Wild
Fanghua Yu, Jinjin Gu, Zheyuan Li +6
We introduce SUPIR (Scaling-UP Image Restoration), a groundbreaking image restoration method that harnesses generative prior and the power of model scaling up. Leveraging multi-mod…