Search papers, labs, and topics across Lattice.
The paper introduces UCAN, a lightweight neural network architecture for image super-resolution that unifies convolution and attention mechanisms to efficiently expand the receptive field. UCAN employs window-based spatial attention, Hedgehog Attention, and a distillation-based large-kernel module to capture both local textures and long-range dependencies while minimizing computational cost. Experiments demonstrate that UCAN achieves state-of-the-art performance on Manga109 and BSDS100 datasets with significantly fewer MACs compared to existing lightweight super-resolution models.
UCAN achieves state-of-the-art image super-resolution with significantly fewer computations, proving that expansive receptive fields don't require heavyweight architectures.
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 ($4\times$), 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.
