PocketSR: The Super-Resolution Expert in Your Pocket Mobiles
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912626281283584 |
|---|---|
| author | Sun, Haoze Jiang, Linfeng Li, Fan Pei, Renjing Wang, Zhixin Guo, Yong Xu, Jiaqi Chen, Haoyu Han, Jin Song, Fenglong Yang, Yujiu Li, Wenbo |
| author_facet | Sun, Haoze Jiang, Linfeng Li, Fan Pei, Renjing Wang, Zhixin Guo, Yong Xu, Jiaqi Chen, Haoyu Han, Jin Song, Fenglong Yang, Yujiu Li, Wenbo |
| contents | Real-world image super-resolution (RealSR) aims to enhance the visual quality of in-the-wild images, such as those captured by mobile phones. While existing methods leveraging large generative models demonstrate impressive results, the high computational cost and latency make them impractical for edge deployment. In this paper, we introduce PocketSR, an ultra-lightweight, single-step model that brings generative modeling capabilities to RealSR while maintaining high fidelity. To achieve this, we design LiteED, a highly efficient alternative to the original computationally intensive VAE in SD, reducing parameters by 97.5% while preserving high-quality encoding and decoding. Additionally, we propose online annealing pruning for the U-Net, which progressively shifts generative priors from heavy modules to lightweight counterparts, ensuring effective knowledge transfer and further optimizing efficiency. To mitigate the loss of prior knowledge during pruning, we incorporate a multi-layer feature distillation loss. Through an in-depth analysis of each design component, we provide valuable insights for future research. PocketSR, with a model size of 146M parameters, processes 4K images in just 0.8 seconds, achieving a remarkable speedup over previous methods. Notably, it delivers performance on par with state-of-the-art single-step and even multi-step RealSR models, making it a highly practical solution for edge-device applications. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_03012 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | PocketSR: The Super-Resolution Expert in Your Pocket Mobiles Sun, Haoze Jiang, Linfeng Li, Fan Pei, Renjing Wang, Zhixin Guo, Yong Xu, Jiaqi Chen, Haoyu Han, Jin Song, Fenglong Yang, Yujiu Li, Wenbo Computer Vision and Pattern Recognition Real-world image super-resolution (RealSR) aims to enhance the visual quality of in-the-wild images, such as those captured by mobile phones. While existing methods leveraging large generative models demonstrate impressive results, the high computational cost and latency make them impractical for edge deployment. In this paper, we introduce PocketSR, an ultra-lightweight, single-step model that brings generative modeling capabilities to RealSR while maintaining high fidelity. To achieve this, we design LiteED, a highly efficient alternative to the original computationally intensive VAE in SD, reducing parameters by 97.5% while preserving high-quality encoding and decoding. Additionally, we propose online annealing pruning for the U-Net, which progressively shifts generative priors from heavy modules to lightweight counterparts, ensuring effective knowledge transfer and further optimizing efficiency. To mitigate the loss of prior knowledge during pruning, we incorporate a multi-layer feature distillation loss. Through an in-depth analysis of each design component, we provide valuable insights for future research. PocketSR, with a model size of 146M parameters, processes 4K images in just 0.8 seconds, achieving a remarkable speedup over previous methods. Notably, it delivers performance on par with state-of-the-art single-step and even multi-step RealSR models, making it a highly practical solution for edge-device applications. |
| title | PocketSR: The Super-Resolution Expert in Your Pocket Mobiles |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2510.03012 |