FoundIR-v2：为图像复原基础模型优化预训练数据配比
FoundIR-v2: Optimizing Pre-Training Data Mixtures for Image Restoration Foundation Model
把「不同复原任务的数据配比」当作可优化的变量，用数据混合律做动态调度，并用 MoE 调度器为每个任务分配任务自适应的扩散先验；做 all-in-one 复原或多任务联合训练时，可直接借鉴这套数据均衡调度与专家路由思路。
原文依据 adopts a data equilibrium scheduling paradigm to dynamically optimize the proportions of mixed training datasets from different tasks
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论近期研究显示，预训练数据规模与质量的提升推动了图像复原基础模型的显著进展。在本工作中，我们发现来自不同复原任务的数据混合比例同样是直接决定一体化图像复原模型整体性能的关键因素。为此，我们提出一个高容量、基于扩散的图像复原基础模型 FoundIR-v2，它采用数据均衡调度范式，动态优化来自不同任务的混合训练数据集的比例。借助数据混合律，我们的方法保证了数据集构成的均衡，使模型能够在多种任务上取得一致的泛化性与全面的性能。此外，我们在生成式预训练中引入了有效的混合专家（MoE）驱动调度器，为每个复原任务灵活分配任务自适应的扩散先验，以顾及不同任务所呈现的退化形式与程度的差异。大量实验表明，我们的方法能够应对更广泛真实场景中的 50 余个子任务，并相较最优方法取得了良好性能。
英文原文摘要 1274 字符
Recent studies have witnessed significant advances in image restoration foundation models driven by improvements in the scale and quality of pre-training data. In this work, we find that the data mixture proportions from different restoration tasks are also a critical factor directly determining the overall performance of all-in-one image restoration models. To this end, we propose a high-capacity diffusion-based image restoration foundation model, FoundIR-v2, which adopts a data equilibrium scheduling paradigm to dynamically optimize the proportions of mixed training datasets from different tasks. By leveraging the data mixing law, our method ensures a balanced dataset composition, enabling the model to achieve consistent generalization and comprehensive performance across diverse tasks. Furthermore, we introduce an effective Mixture-of-Experts (MoE)-driven scheduler into generative pre-training to flexibly allocate task-adaptive diffusion priors for each restoration task, accounting for the distinct degradation forms and levels exhibited by different tasks. Extensive experiments demonstrate that our method can address over 50 sub-tasks across a broader scope of real-world scenarios and achieves favorable performance against state-of-the-art approaches.