UCAN：用于轻量级超分辨率中扩展感受野的统一卷积注意力网络
UCAN: Unified Convolutional Attention Network for Expansive Receptive Fields in Lightweight Super-Resolution
UCAN 通过窗口注意力与 Hedgehog Attention 结合，并引入蒸馏式大核模块，在极低计算量下扩大有效感受野，为轻量级超分辨率及其他图像复原任务提供了可复用的高效架构设计。
原文依据 UCAN combines window-based spatial attention with a Hedgehog Attention mechanism to model both local texture and long-range dependencies
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论混合 CNN-Transformer 架构在图像超分辨率中取得了显著成果，但扩展注意力窗口或卷积核会大幅增加计算成本，限制了在资源受限设备上的部署。我们提出 UCAN，一种统一卷积与注意力以高效扩展有效感受野的轻量级网络。UCAN 将基于窗口的空间注意力与 Hedgehog 注意力机制相结合，以建模局部纹理和长程依赖，并引入基于蒸馏的大核模块，在不增加大量计算的情况下保留高频结构。此外，我们采用跨层参数共享以进一步降低复杂度。在 Manga109（4 倍）上，UCAN-L 仅以 48.4G MACs 达到 31.63 dB 峰值信噪比，超越了近期轻量级模型。在 BSDS100 上，UCAN 达到 27.79 dB，优于模型规模显著更大的方法。大量实验表明，UCAN 在精度、效率和可扩展性之间实现了优越的权衡，非常适合实际的高分辨率图像复原。
英文原文摘要 1092 字符
Hybrid CNN-Transformer architectures achieve strong results in image super-resolution, but scaling attention windows or convolution kernels significantly increases computational cost, limiting deployment on resource-constrained devices. We present UCAN, a lightweight network that unifies convolution and attention to expand the effective receptive field efficiently. UCAN combines window-based spatial attention with a Hedgehog Attention mechanism to model both local texture and long-range dependencies, and introduces a distillation-based large-kernel module to preserve high-frequency structure without heavy computation. In addition, we employ cross-layer parameter sharing to further reduce complexity. On Manga109 (4x), UCAN-L achieves 31.63 dB PSNR with only 48.4G MACs, surpassing recent lightweight models. On BSDS100, UCAN attains 27.79 dB, outperforming methods with significantly larger models. Extensive experiments show that UCAN achieves a superior trade-off between accuracy, efficiency, and scalability, making it well-suited for practical high-resolution image restoration.