Multi-scale Attention Network for Single Image Super-Resolution (2209.14145v3)
Abstract: ConvNets can compete with transformers in high-level tasks by exploiting larger receptive fields. To unleash the potential of ConvNet in super-resolution, we propose a multi-scale attention network (MAN), by coupling classical multi-scale mechanism with emerging large kernel attention. In particular, we proposed multi-scale large kernel attention (MLKA) and gated spatial attention unit (GSAU). Through our MLKA, we modify large kernel attention with multi-scale and gate schemes to obtain the abundant attention map at various granularity levels, thereby aggregating global and local information and avoiding potential blocking artifacts. In GSAU, we integrate gate mechanism and spatial attention to remove the unnecessary linear layer and aggregate informative spatial context. To confirm the effectiveness of our designs, we evaluate MAN with multiple complexities by simply stacking different numbers of MLKA and GSAU. Experimental results illustrate that our MAN can perform on par with SwinIR and achieve varied trade-offs between state-of-the-art performance and computations.
- NTIRE 2017 challenge on single image super-resolution: Dataset and study. In CVPRW, pages 1122–1131, Honolulu, USA, 2017. IEEE Computer Society.
- Low-complexity single-image super-resolution based on nonnegative neighbor embedding. In BMVC, pages 1–10, Surrey, UK, 2012. BMVA Press.
- Equivalent transformation and dual stream network construction for mobile image super-resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 14102–14111, 2023.
- Pre-trained image processing transformer. In CVPR, pages 12299–12310, virtual, 2021. Computer Vision Foundation / IEEE.
- Simple baselines for image restoration. arXiv preprint arXiv:2204.04676, 2022a.
- Activating more pixels in image super-resolution transformer. arXiv preprint arXiv:2205.04437, 2022b.
- Dual aggregation transformer for image super-resolution. In Proceedings of the IEEE/CVF international conference on computer vision, pages 12312–12321, 2023.
- Second-order attention network for single image super-resolution. In CVPR, pages 11065–11074, Long Beach, USA, 2019. Computer Vision Foundation / IEEE.
- Language modeling with gated convolutional networks. In ICML, pages 933–941, Sydney, Australia, 2017. PMLR.
- Imagenet: A large-scale hierarchical image database. In CVPR, pages 248–255, Miami, USA, 2009. IEEE Computer Society.
- Image super-resolution using deep convolutional networks. IEEE TPAMI, 38(2):295–307, 2016a.
- Accelerating the super-resolution convolutional neural network. In ECCV, pages 391–407, Amsterdam, The Netherlands, 2016b. Springer.
- An image is worth 16x16 words: Transformers for image recognition at scale. In ICLR, Virtual Event, Austria, 2021.
- Interpreting super-resolution networks with local attribution maps. In CVPR, pages 9199–9208, 2021.
- Visual attention network. arXiv preprint arXiv:2202.09741, 2022.
- Single image super-resolution using gaussian process regression. In CVPR, pages 449–456, Colorado Springs, USA, 2011.
- Transformer quality in linear time. arXiv preprint arXiv:2202.10447, 2022.
- Single image super-resolution from transformed self-exemplars. In CVPR, pages 5197–5206, Boston, MA, USA, 2015. IEEE Computer Society.
- Lightweight image super-resolution with information multi-distillation network. In ACM MM, pages 2024–2032, Nice, France, 2019. ACM.
- Hierarchical dense recursive network for image super-resolution. PR, 107:107475, 2020.
- Deeply-recursive convolutional network for image super-resolution. In CVPR, pages 1637–1645, Las Vegas, USA, 2016a. IEEE Computer Society.
- Accurate image super-resolution using very deep convolutional networks. In CVPR, pages 1646–1654, Las Vegas, NV, USA, 2016b. IEEE Computer Society.
- Adam: A method for stochastic optimization. In ICLR, San Diego, USA, 2015.
- Deep laplacian pyramid networks for fast and accurate super-resolution. In CVPR, pages 5835–5843, Honolulu, USA, 2017. IEEE Computer Society.
- Multi-scale residual network for image super-resolution. In ECCV, pages 527–542, Munich, Germany, 2018. Springer.
- LAPAR: linearly-assembled pixel-adaptive regression network for single image super-resolution and beyond. In NeurIPS, virtual, 2020.
- On efficient transformer and image pre-training for low-level vision. arXiv preprint arXiv:2112.10175, 2021.
- Efficient and explicit modelling of image hierarchies for image restoration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18278–18289, 2023a.
- Lsdir: A large scale dataset for image restoration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1775–1787, 2023b.
- Swinir: Image restoration using swin transformer. In ICCVW, pages 1833–1844, Montreal, Canada, 2021. IEEE.
- Enhanced deep residual networks for single image super-resolution. In CVPRW, pages 1132–1140. IEEE Computer Society, 2017.
- Revisiting rcan: Improved training for image super-resolution. arXiv preprint arXiv:2201.11279, 2022.
- Residual feature aggregation network for image super-resolution. In CVPR, pages 2356–2365, Seattle, USA, 2020. Computer Vision Foundation / IEEE.
- Swin transformer: Hierarchical vision transformer using shifted windows. In ICCV, pages 9992–10002, Montreal, Canada, 2021. IEEE.
- A convnet for the 2020s. arXiv preprint arXiv:2201.03545, 2022.
- Latticenet: Towards lightweight image super-resolution with lattice block. In ECCV, pages 272–289, Glasgow, UK, 2020. Springer.
- Dynamic high-pass filtering and multi-spectral attention for image super-resolution. In ICCV, pages 4268–4277, Montreal, Canada, 2021. IEEE.
- A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics. In ICCV, pages 416–425, Vancouver, Canada, 2001. IEEE Computer Society.
- Sketch-based manga retrieval using manga109 dataset. Multim. Tools Appl., 76(20):21811–21838, 2017.
- Image super-resolution with non-local sparse attention. In CVPR, pages 3517–3526, virtual, 2021. Computer Vision Foundation / IEEE.
- Single image super-resolution via a holistic attention network. In ECCV, pages 191–207, Glasgow, UK, 2020. Springer.
- Pytorch: An imperative style, high-performance deep learning library. In NeurIPS, pages 8024–8035, Vancouver, Canada, 2019.
- Image super-resolution using gradient profile prior. In CVPR, pages 1–8, Anchorage, USA, 2008.
- Shufflemixer: An efficient convnet for image super-resolution. arXiv preprint arXiv:2205.15175, 2022.
- Image super-resolution via deep recursive residual network. In CVPR, pages 2790–2798, Honolulu, HI, USA, 2017. IEEE Computer Society.
- Super resolution using edge prior and single image detail synthesis. In CVPR, pages 2400–2407, San Francisco, USA, 2010.
- PVT v2: Improved baselines with pyramid vision transformer. Comput. Vis. Media, 8(3):415–424, 2022.
- Yan Wang. Edge-enhanced feature distillation network for efficient super-resolution. In CVPRW, pages 777–785, 2022.
- Camixersr: Only details need more “attention". arXiv preprint arXiv:2402.19289, 2024.
- Image quality assessment: from error visibility to structural similarity. IEEE TIP, 13(4):600–612, 2004.
- Efficient non-local contrastive attention for image super-resolution. arXiv preprint arXiv:2201.03794, 2022.
- Dipnet: Efficiency distillation and iterative pruning for image super-resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1692–1701, 2023.
- Metaformer is actually what you need for vision. arXiv preprint arXiv:2111.11418, 2021.
- See more details: Efficient image super-resolution by experts mining. arXiv preprint arXiv:2402.03412, 2024.
- On single image scale-up using sparse-representations. In Curves and Surfaces - 7th International Conference, pages 711–730, Avignon, France, 2010. Springer.
- Edge-oriented convolution block for real-time super resolution on mobile devices. In ACM MM, pages 4034–4043, Virtual Event, China, 2021. ACM.
- Efficient long-range attention network for image super-resolution. arXiv preprint arXiv:2203.06697, 2022.
- Image super-resolution using very deep residual channel attention networks. In ECCV, pages 294–310, Munich, Germany, 2018. Springer.
- Cross-scale internal graph neural network for image super-resolution. In NeurIPS, virtual, 2020.
- Yan Wang (733 papers)
- Yusen Li (11 papers)
- Gang Wang (407 papers)
- Xiaoguang Liu (19 papers)