通过数据增强与测试时自集成，显著提升图像去噪性能。
Beyond Model Design: Data-Centric Training and Self-Ensemble for Gaussian Color Image Denoising
- 用更大更丰富的数据集分两阶段训练，强化模型泛化能力。
- 在验证集上达30.762 dB PSNR，比基线提升3.366 dB。
- 适合关注实际应用中去噪性能优化的研究者与开发者。
本文针对NTIRE 2026图像去噪挑战（高斯彩色图像去噪，噪声水平σ=50）提出解决方案。未设计新模型架构，而是从数据驱动训练和测试时能力释放两个方向优化成熟的Restormer结构。基于公开的Restormer σ=50预训练基线，扩展多数据集训练，引入更大更多样化的公共图像语料库，并采用两阶段优化策略。推理时采用×8几何自集成进一步释放模型潜力。保留TLC风格局部推理封装以保证实现一致性；但系统消融实验表明其贡献可忽略。在包含100张图像的挑战验证集上，最终提交结果达到30.762 dB PSNR和0.861 SSIM，相比基线最高提升3.366 dB PSNR。消融分析显示，性能提升主要来自扩展训练数据和两阶段优化，自集成带来微弱但持续的增益。
原文摘要 · Abstract (English)
This paper presents our solution to the NTIRE 2026 Image Denoising Challenge (Gaussian color image denoising at fixed noise level $σ= 50$). Rather than proposing a new restoration backbone, we revisit the performance boundary of the mature Restormer architecture from two complementary directions: stronger data-centric training and more complete Test-Time capability release. Starting from the public Restormer $σ\!=\!50$ baseline, we expand the standard multi-dataset training recipe with larger and more diverse public image corpora and organize optimization into two stages. At inference, we apply $\times 8$ geometric self-ensemble to further release model capacity. A TLC-style local inference wrapper is retained for implementation consistency; however, systematic ablation reveals its quantitative contribution to be negligible in this setting. On the challenge validation set of 100 images, our final submission achieves 30.762 dB PSNR and 0.861 SSIM, improving over the public Restormer $σ\!=\!50$ pretrained baseline by up to 3.366 dB PSNR. Ablation studies show that the dominant gain originates from the expanded training corpus and the two-stage optimization schedule, and self-ensemble provides marginal but consistent improvement.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。