Two-Stage Vision Transformer for Image Restoration: Colorization Pretraining + Residual Upsampling
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866912745575677952 |
|---|---|
| author | Chaudhary, Aditya Singh, Prachet Dev Jha, Ankit |
| author_facet | Chaudhary, Aditya Singh, Prachet Dev Jha, Ankit |
| contents | In computer vision, Single Image Super-Resolution (SISR) is still a difficult problem. We present ViT-SR, a new technique to improve the performance of a Vision Transformer (ViT) employing a two-stage training strategy. In our method, the model learns rich, generalizable visual representations from the data itself through a self-supervised pretraining phase on a colourization task. The pre-trained model is then adjusted for 4x super-resolution. By predicting the addition of a high-frequency residual image to an initial bicubic interpolation, this design simplifies residual learning. ViT-SR, trained and evaluated on the DIV2K benchmark dataset, achieves an impressive SSIM of 0.712 and PSNR of 22.90 dB. These results demonstrate the efficacy of our two-stage approach and highlight the potential of self-supervised pre-training for complex image restoration tasks. Further improvements may be possible with larger ViT architectures or alternative pretext tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_02512 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Two-Stage Vision Transformer for Image Restoration: Colorization Pretraining + Residual Upsampling Chaudhary, Aditya Singh, Prachet Dev Jha, Ankit Computer Vision and Pattern Recognition In computer vision, Single Image Super-Resolution (SISR) is still a difficult problem. We present ViT-SR, a new technique to improve the performance of a Vision Transformer (ViT) employing a two-stage training strategy. In our method, the model learns rich, generalizable visual representations from the data itself through a self-supervised pretraining phase on a colourization task. The pre-trained model is then adjusted for 4x super-resolution. By predicting the addition of a high-frequency residual image to an initial bicubic interpolation, this design simplifies residual learning. ViT-SR, trained and evaluated on the DIV2K benchmark dataset, achieves an impressive SSIM of 0.712 and PSNR of 22.90 dB. These results demonstrate the efficacy of our two-stage approach and highlight the potential of self-supervised pre-training for complex image restoration tasks. Further improvements may be possible with larger ViT architectures or alternative pretext tasks. |
| title | Two-Stage Vision Transformer for Image Restoration: Colorization Pretraining + Residual Upsampling |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2512.02512 |