Learning A Single Network for Scale-Arbitrary Super-Resolution
This addresses the problem of flexible image super-resolution for applications requiring non-standard scaling, though it is incremental as it builds on existing networks.
The paper tackles the limitation of existing super-resolution networks that only work with specific integer scales by proposing a plug-in module to enable scale-arbitrary super-resolution, achieving promising results for non-integer and asymmetric scales while maintaining state-of-the-art performance for integer scales with minimal additional computational cost.
Recently, the performance of single image super-resolution (SR) has been significantly improved with powerful networks. However, these networks are developed for image SR with a single specific integer scale (e.g., x2;x3,x4), and cannot be used for non-integer and asymmetric SR. In this paper, we propose to learn a scale-arbitrary image SR network from scale-specific networks. Specifically, we propose a plug-in module for existing SR networks to perform scale-arbitrary SR, which consists of multiple scale-aware feature adaption blocks and a scale-aware upsampling layer. Moreover, we introduce a scale-aware knowledge transfer paradigm to transfer knowledge from scale-specific networks to the scale-arbitrary network. Our plug-in module can be easily adapted to existing networks to achieve scale-arbitrary SR. These networks plugged with our module can achieve promising results for non-integer and asymmetric SR while maintaining state-of-the-art performance for SR with integer scale factors. Besides, the additional computational and memory cost of our module is very small.