DRNet: All-in-One Image Restoration via Prior-Guided Dynamic Reparameterization
Ao Li, Xiaoning Liu, Sheng Li, Yapeng Du, Zhen Long, Lei Luo, Le Zhang, Ce Zhu
arxiv.org/abs/2605.08627 [𝚌𝚜.𝙲𝚅]
💬Accepted by IEEE TMM
ALT All-in-one image restoration aims to handle diverse degradations within a single model. However, existing methods often suffer from three key limitations: 1) per-input computational overhead from dynamic degradation estimation; 2) optimization challenges due to task heterogeneity; and 3) inefficient, frequency-agnostic encoder designs. To overcome these, we introduce the Dynamic Reparameterization Network (DRNet), a novel framework operating on an initialization-stage reconfiguration paradigm that fundamentally eliminates per-input overhead. At its core, a Dynamic Reparameterization MLP (DRMLP) guided by a Task-Specific Modulator (TSM), which effectively mitigates task heterogeneity by orchestrating both specific restoration goals and a versatile general-purpose mode within a unified architecture. Furthermore, we incorporate a Continuous Wavelet Transform Encoder (CWTE) that explicitly leverages frequency characteristics via wavelet decomposition for a lightweight yet powerful design.
@Connect2NonStop yes, the Nov - Dec issue of The Connection is now live. In this issue you will read:
For the Nonstop community, we once again bore witness to the inroads the latest Nonstop Compute product lines were making into the Nonstop customer base. The latest edition to the product family of Nonstop Compute converged systems, the Nonstop NS9 X5 and the Nonstop NS5 X5, has been well received with enterprises across all major regions taking delivery of these latest iterations of an X86-architected system ...
... “We move the data that moves your business,” it was just as appropriate a venue to promote our M&Ms strategy. More than just a reference to a chocolate treat but an acknowledgement that with the introduction of its latest company tag line, “We move the data that moves your business.” for NTI, it’s all about Movement, Modernization, Migration and the benefits that come with deploying DRNet®/Unified at the center of your data movement strategy.
And this is just the opening and to read more of what the NTI team brought to the community apart from M&Ms, check out their latest article:
connect2nonstop.com/for-nti-…
DRNET-X: a deep learning-based framework for diabetic retinopathy image detection using hybrid preprocessing and CNN-based classification
link.springer.com/article/10…@grok, pode resumir?
5/ Results: our NuDRNet (DR-Learner/DRNet variant) and NuSNet (representation-based SNet variant) outperform baselines and, strikingly, almost match the oracle-propensity performance in several settings.