Efficient Concertormer for Image Deblurring and Beyond
Abstract
The Transformer architecture has excelled in NLP and vision tasks, but its self-attention complexity grows quadratically with image size, making high-resolution tasks computationally expensive. We introduce Concertormer, featuring Concerto Self-Attention (CSA) for image deblurring. CSA splits self-attention into global and local components while retaining partial information in additional dimensions, achieving linear complexity. A Cross-Dimensional Communication module enhances expressiveness by linearly combining attention maps. Additionally, our gated-dconv MLP merges the two-staged Transformer design into a single stage. Extensive evaluations show our method performs favorably against state-of-the-art works in deblurring, deraining, and JPEG artifact removal.
Cite
Text
Kuo et al. "Efficient Concertormer for Image Deblurring and Beyond." International Conference on Computer Vision, 2025. doi:10.1109/ICCV51701.2025.01361Markdown
[Kuo et al. "Efficient Concertormer for Image Deblurring and Beyond." International Conference on Computer Vision, 2025.](https://mlanthology.org/iccv/2025/kuo2025iccv-efficient/) doi:10.1109/ICCV51701.2025.01361BibTeX
@inproceedings{kuo2025iccv-efficient,
title = {{Efficient Concertormer for Image Deblurring and Beyond}},
author = {Kuo, Pin-Hung and Pan, Jinshan and Chien, Shao-Yi and Yang, Ming-Hsuan},
booktitle = {International Conference on Computer Vision},
year = {2025},
pages = {14665-14675},
doi = {10.1109/ICCV51701.2025.01361},
url = {https://mlanthology.org/iccv/2025/kuo2025iccv-efficient/}
}