Abstract
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.
Abstract (translated)
在计算机视觉领域，单图像超分辨率（SISR）仍然是一个具有挑战性的问题。我们提出了一种名为ViT-SR的新技术，该技术通过两阶段训练策略来提高Vision Transformer (ViT) 的性能。我们的方法中，模型通过在着色任务上的自我监督预训练阶段从数据本身学习丰富的、可泛化的视觉表示。然后将预训练的模型调整为进行4倍超分辨率处理。通过预测向初始双三次插值图像添加高频残差图像，这种设计简化了残差学习的过程。 在DIV2K基准数据集上进行训练和评估后，ViT-SR取得了令人印象深刻的SSIM（结构相似性指数）0.712 和 PSNR（峰值信噪比）22.90 dB 的结果。这些成果展示了我们两阶段方法的有效性，并突显了自我监督预训练在复杂图像恢复任务中的潜力。通过使用更大规模的ViT架构或替代前置任务，可能进一步改进性能。
URL
https://arxiv.org/abs/2512.02512
