DeblurDiNAT: A Lightweight and Effective Transformer for Image Deblurring

arXiv:2403.13163v1 Announce Type: new
Abstract: Blurry images may contain local and global non-uniform artifacts, which complicate the deblurring process and make it more challenging to achieve satisfactory results. Recently, Transformers generate improved deblurring outcomes than existing CNN architectures. However, the large model size and long inference time are still two bothersome issues which have not been fully explored. To this end, we propose DeblurDiNAT, a compact encoder-decoder Transformer which efficiently restores clean images from real-world blurry ones. We adopt an …

abstract architectures arxiv cnn cs.cv generate global however image images inference process results transformer transformers type uniform