Jun 3, 2021 · Extensive experiments demonstrate that our proposed C2-Matching significantly outperforms state of the arts by over 1dB on the standard CUFED5 ...

This repository contains the implementation of the following paper: Robust Reference-based Super-Resolution via C2-Matching.

Extensive experiments demonstrate that our proposed C2-Matching significantly outperforms state of the arts by over 1dB on the standard. CUFED5 benchmark.

The proposed C2-Matching significantly outperforms state of the arts by over 1dB on the standard CUFED5 benchmark and shows great generalizability on WR-SR ...

Extensive experiments demonstrate that our proposed C2-Matching significantly outperforms state of the arts by over 1dB on the standard CUFED5 benchmark.

In this supplementary file, we will explain the network structures (i.e. Contrastive Correspondence Network and. Restoration Network) and training details ...

The teacher-student correlation distillation enables the contrastive correspondence network to compute correspondences more accurately for texture regions.

Dec 19, 2022 · We propose C2-Matching in this work, which performs explicit robust matching crossing transformation and resolution.

The correspondence matching module is based on C2-Matching [22] , which only uses features from rendered and reference images. We use the VGG extractorr [40] to ...

To tackle these challenges, we propose C2-Matching in this work, which performs explicit robust matching crossing transformation and resolution. 1.