Celebrating Samsung R&D Institute China-Beijing's Outstanding Achievements at NTIRE at the CVPR 2026

The New Trends in Image Restoration and Enhancement (NTIRE 2026) workshop, held alongside the Conference on Computer Vision and Pattern Recognition (CVPR 2026), served as a key venue for sharing the latest advances in image processing and restoration. Samsung R&D Institute China-Beijing (SRC-B) participated in multiple challenges and earned top-tier results across several tracks. This remarkable performance highlights SRC-B's relentless pursuit of innovation and its growing leadership in the field of image restoration and enhancement.

1. Outstanding Achievements in NTIRE at the CVPR 2026

1) Image Super-Resolution (x4) Challenge - Top 1

SRC-B secured the first place by introducing a novel cascaded framework that balances global structural coherence with local texture fidelity. The proposed network integrates three key modules: the Global Optimization Module (GOM), based on HAT to capture long-range spatial dependencies; the Detail Enhancement Module (DEM), based on NAFNet to restore high-frequency information; and the Dynamic Fusion Module (DFM) for adaptive feature aggregation. A critical innovation is the Semantic Injection Module (SIM) that bridges global and local learning by selectively injecting deep semantic features into intermediate DEM layers.

2) Blind Computational Aberration Correction Challenge - Top 1

SRC-B achieved first place in this inaugural challenge by developing a dual-branch hybrid architecture that complements physics-aware restoration with data-driven restoration. The solution leverages SwinIR with a point spread function (PSF) decoder as the physics-aware branch and Nonlinear Activation Free Network (NAFNet) as the data-driven branch, connected through an innovative Adaptive Fusion Module with attention mechanisms. This design dynamically suppresses unreliable features while amplifying consistent signals from both branches, ensuring robust performance across diverse aberration types.

3) Mobile Real-World Image Super-Resolution Challenge - Top 3

SRC-B also achieved third place by balancing image quality with device-side efficiency through extensive model pruning and optimization work on mobile hardware.

2. Technical Excellence and Innovation

Innovative architectural designs and meticulous problem-solving strategies Drove SRC-B’s success. Key innovations include:

Cascaded Framework with Semantic Injection: Groundbreaking architecture separating global structure reconstruction from local detail enhancement through intelligent semantic injection for super-resolution.

Hybrid Physics-Aware and Data-Driven Architecture: Pioneering dual-branch system seamlessly integrating physics-based and data-driven approaches for aberration correction.

Progressive Multistage-Training Strategy: Sophisticated protocols allowing independent convergence before cooperative fusion learning.

Data Augmentation and Progressive Learning: Innovative augmentation techniques combined with gradual patch size increases for consistent performance improvements.

3. Prospects and Commercialization

SRC-B’s achievements hold significant potential for real-world applications. The developed technologies are well-positioned for commercialization in areas such as:

Mobile Camera ISP: Enhanced super-resolution and aberration correction capabilities improving smartphone camera performance, especially on devices with compact camera modules.

Computational Imaging and VR/AR: Aberration correction revolutionizing virtual and augmented reality displays with sharper, more immersive visual experiences.

Digital Zoom and Professional Photography: Super-resolution technologies enabling high-fidelity images at high magnifications without artifacts.

SRC-B’s remarkable performance in NTIRE at the CVPR 2026 stands as a testament to the team’s dedication, innovation, and technical excellence. By tackling complex challenges and consistently pushing the boundaries of image restoration, SRC-B is setting new benchmarks in the field. As these innovative technologies move from research into real-world applications, they hold the potential to transform how we capture and enhance visual information, further reinforcing SRC-B’s growing leadership in the industry.