Multi-Scale Implicit Transformer with Re-parameterize for Arbitrary-Scale Super-Resolution

arXiv:2403.06536v1 Announce Type: new
Abstract: Recently, the methods based on implicit neural representations have shown excellent capabilities for arbitrary-scale super-resolution (ASSR). Although these methods represent the features of an image by generating latent codes, these latent codes are difficult to adapt for different magnification factors of super-resolution, which seriously affects their performance. Addressing this, we design Multi-Scale Implicit Transformer (MSIT), consisting of an Multi-scale Neural Operator (MSNO) and Multi-Scale Self-Attention (MSSA). Among them, MSNO obtains multi-scale latent codes through feature …

abstract adapt arxiv capabilities cs.cv features image implicit neural representations performance scale transformer type