NTIRE Workshop & Challenges @ CVPR 2026 • Denver, Colorado
Single-Image Super-Resolution (×4) Challenge
In application scenarios such as mobile imaging, medical imaging, and satellite remote sensing, low-resolution images often lose details due to hardware limitations and imaging conditions, affecting subsequent analysis and practical use. This track focuses on 4× super-resolution reconstruction (×4 SR), encouraging participants to restore rich texture and structural information of high-resolution images from low-resolution inputs.
Challenge Overview
This challenge is part of the NTIRE Workshop & Challenges co-located with CVPR 2026. NTIRE (New Trends in Image Restoration and Enhancement) is one of the most influential international competitions and workshops in the field of image restoration and enhancement, continuously promoting fair comparison and continuous progress of image restoration algorithms under unified data and evaluation protocols.
Why super-resolution is needed
With the continuous development of computer vision, single-image super-resolution (Single-Image Super-Resolution, ×4) has become one of the key directions in the field of image restoration. In application scenarios such as mobile imaging, medical imaging, and satellite remote sensing, low-resolution images often lose details due to hardware limitations and imaging conditions, affecting subsequent analysis and practical use.
Challenge goal
This track focuses on 4× super-resolution reconstruction (×4 SR), encouraging participants to restore rich texture and structural information of high-resolution images from low-resolution inputs, promoting the usability and innovation of algorithms in real-world scenarios.
Data & Competition Phases
Where to get the data
All datasets are hosted on the competition platform under the Files section.
Development Phase
Training set (with input-output pairs) and validation set (input only) are released. Participants can train models based on the training set and debug and iteratively optimize through the validation set.
Test Phase
After the final test set (input only) is released, participants submit super-resolution results from model outputs, which are uniformly evaluated by the organizing committee to give final rankings.
Evaluation
Dual-track evaluation
The goal of this challenge is to develop advanced single-image super-resolution algorithms that reconstruct high-quality high-resolution images from low-resolution inputs based on example data, restoring rich details and texture information as much as possible. The competition is divided into dual-track evaluation:
- Restoration Quality: Using PSNR as the core metric to measure pixel-level restoration accuracy.
- Perception Quality: Using Perception Score as the core metric, which is obtained by weighting 6 perceptual metrics (LPIPS, DISTS, NIQE, ManIQA, MUSIQ, CLIP-IQA) to evaluate the visual realism and perceptual quality of results.
Submission requirements
- Submit super-resolution results for the test set by the result deadline.
- Submit a Fact Sheet and reproducible code/executable by the code deadline.
- Top-ranked teams will be invited to present at the NTIRE workshop.
Important Dates
Awards & Opportunities
Awards
Top-ranked teams will be invited to publish up to 8-page papers at the NTIRE Workshop, included in the CVPR 2025 Workshop proceedings.
Dual-track independent awards: Restoration Quality and Perception Quality tracks each set up awards, with the top three teams receiving challenge certificates.
Other opportunities
Details about sponsor rewards and travel grants will be updated on the NTIRE website.
Join the Challenge
Competition portal
Register, download data, and submit results on the CodaBench competition page.
Open CodaBenchNTIRE 2026
Learn more about the NTIRE 2026 workshop and other challenges on the official NTIRE website.
Open NTIRE 2026Organizers
Organizer list may be updated.
WeChat Group
Scan the QR code below to join the official WeChat group for announcements and discussion.
Note: If the QR code expires, please scan the QR code at the bottom of the competition homepage to join the group.
