Główne pojęcia
Introducing MambaIR as a simple but effective baseline for image restoration, enhancing global receptive fields and computational efficiency.
Statystyki
"MambaIR outperforms SwinIR by up to 0.45dB on image SR."
"Model size comparisons: Our MambaIR has 16.7M parameters and 439G MACs."
Cytaty
"Despite possessing many attractive properties, there exists an inherent choice dilemma between global receptive fields and efficient computation for current image restoration backbones."
"Extensive experiments demonstrate the superiority of our method, for example, MambaIR outperforms SwinIR by up to 0.45dB on image SR."