Residual trio feature network for efficient super-resolution
Abstract Deep learning-based approaches have demonstrated impressive performance in single-image super-resolution (SISR). Efficient super-resolution compromises the reconstructed image’s quality to have fewer parameters and Flops. Ensured efficiency in image reconstruction and improved reconstructio...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2024-11-01
|
| Series: | Complex & Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s40747-024-01624-8 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832571158044606464 |
|---|---|
| author | Junfeng Chen Mao Mao Azhu Guan Altangerel Ayush |
| author_facet | Junfeng Chen Mao Mao Azhu Guan Altangerel Ayush |
| author_sort | Junfeng Chen |
| collection | DOAJ |
| description | Abstract Deep learning-based approaches have demonstrated impressive performance in single-image super-resolution (SISR). Efficient super-resolution compromises the reconstructed image’s quality to have fewer parameters and Flops. Ensured efficiency in image reconstruction and improved reconstruction quality of the model are significant challenges. This paper proposes a trio branch module (TBM) based on structural reparameterization. TBM achieves equivalence transformation through structural reparameterization operations, which use a complex network structure in the training phase and convert it to a more lightweight structure in the inference, achieving efficient inference while maintaining accuracy. Based on the TBM, we further design a lightweight version of the enhanced spatial attention mini (ESA-mini) and the residual trio feature block (RTFB). Moreover, the multiple RTFBs are combined to construct the residual trio network (RTFN). Finally, we introduce a localized contrast loss for better applicability to the super-resolution task, which enhances the reconstruction quality of the super-resolution model. Experiments show that the RTFN framework proposed in this paper outperforms other state-of-the-art efficient super-resolution methods in terms of inference speed and reconstruction quality. |
| format | Article |
| id | doaj-art-3ec043687d81461e822e83a7043d3159 |
| institution | Kabale University |
| issn | 2199-4536 2198-6053 |
| language | English |
| publishDate | 2024-11-01 |
| publisher | Springer |
| record_format | Article |
| series | Complex & Intelligent Systems |
| spelling | doaj-art-3ec043687d81461e822e83a7043d31592025-02-02T12:49:56ZengSpringerComplex & Intelligent Systems2199-45362198-60532024-11-0111111210.1007/s40747-024-01624-8Residual trio feature network for efficient super-resolutionJunfeng Chen0Mao Mao1Azhu Guan2Altangerel Ayush3College of Artificial Intelligence and Automation, Hohai UniversityCollege of Information Science and Engineering, Hohai UniversityCollege of Information Science and Engineering, Hohai UniversitySchool of ICT, Mongolian University of Science and TechnologyAbstract Deep learning-based approaches have demonstrated impressive performance in single-image super-resolution (SISR). Efficient super-resolution compromises the reconstructed image’s quality to have fewer parameters and Flops. Ensured efficiency in image reconstruction and improved reconstruction quality of the model are significant challenges. This paper proposes a trio branch module (TBM) based on structural reparameterization. TBM achieves equivalence transformation through structural reparameterization operations, which use a complex network structure in the training phase and convert it to a more lightweight structure in the inference, achieving efficient inference while maintaining accuracy. Based on the TBM, we further design a lightweight version of the enhanced spatial attention mini (ESA-mini) and the residual trio feature block (RTFB). Moreover, the multiple RTFBs are combined to construct the residual trio network (RTFN). Finally, we introduce a localized contrast loss for better applicability to the super-resolution task, which enhances the reconstruction quality of the super-resolution model. Experiments show that the RTFN framework proposed in this paper outperforms other state-of-the-art efficient super-resolution methods in terms of inference speed and reconstruction quality.https://doi.org/10.1007/s40747-024-01624-8Image inpaintingImage super-resolutionRe-parameterization |
| spellingShingle | Junfeng Chen Mao Mao Azhu Guan Altangerel Ayush Residual trio feature network for efficient super-resolution Complex & Intelligent Systems Image inpainting Image super-resolution Re-parameterization |
| title | Residual trio feature network for efficient super-resolution |
| title_full | Residual trio feature network for efficient super-resolution |
| title_fullStr | Residual trio feature network for efficient super-resolution |
| title_full_unstemmed | Residual trio feature network for efficient super-resolution |
| title_short | Residual trio feature network for efficient super-resolution |
| title_sort | residual trio feature network for efficient super resolution |
| topic | Image inpainting Image super-resolution Re-parameterization |
| url | https://doi.org/10.1007/s40747-024-01624-8 |
| work_keys_str_mv | AT junfengchen residualtriofeaturenetworkforefficientsuperresolution AT maomao residualtriofeaturenetworkforefficientsuperresolution AT azhuguan residualtriofeaturenetworkforefficientsuperresolution AT altangerelayush residualtriofeaturenetworkforefficientsuperresolution |