A Hybrid CNN-Transformer Architecture with Frequency Domain Contrastive Learning for Image Deraining

TL;DR

This paper proposes a hybrid CNN-Transformer model utilizing frequency domain contrastive learning to improve the effectiveness of removing rain streaks from images.

Contribution

It introduces a novel hybrid architecture combining CNN and Transformer with frequency domain contrastive learning for image deraining.

Findings

Enhanced deraining performance over existing methods

Improved rain streak removal quality

Better preservation of image details

Abstract

Image deraining is a challenging task that involves restoring degraded images affected by rain streaks.