面孔幻觉研究：数据集偏差的探索性研究
Face Hallucination Revisited: An Exploratory Study on Dataset Bias
摘要：
当代的人脸幻觉（FH）模型表现出相当大的能力，从低分辨率（LR）的人脸图像重建高分辨率（HR）的细节。这种能力通常是从对应的HR-LR图像对的示例中学习到的，该HR-LR图像对是通过对HR地面实况数据进行人工下采样而创建的。该下采样（或降级）过程不仅定义LR训练数据的特性，而且还确定学习的FH模型最终能够处理的图像降级的类型。如果现实世界LR图像遇到的图像特征与训练期间看到的图像特征不同，则FH模型仍有望表现良好，但实际上可能无法产生所需的结果。在本文中，我们研究这个问题，并探讨由训练数据的特点引入FH模型的偏差。我们系统地分析了几种FH模型在各种场景下的泛化能力，其中图像的退化函数与训练设置不匹配，并使用综合降级以及现实生活中的低质量图像进行实验。我们提出了几个有趣的发现，提供深入了解FH模型现有的问题，并指出未来的研究方向。
英文摘要：
Contemporary face hallucination (FH) models exhibit considerable ability to reconstruct high-resolution (HR) details from low-resolution (LR) face images. This ability is commonly learned from examples of corresponding HR-LR image pairs, created by artificially down-sampling the HR ground truth data. This down-sampling (or degradation) procedure not only defines the characteristics of the LR training data, but also determines the type of image degradations the learned FH models are eventually able to handle. If the image characteristics encountered with real-world LR images differ from the ones seen during training, FH models are still expected to perform well, but in practice may not produce the desired results. In this paper we study this problem and explore the bias introduced into FH models by the characteristics of the training data. We systematically analyze the generalization capabilities of several FH models in various scenarios, where the image the degradation function does not match the training setup and conduct experiments with synthetically downgraded as well as real-life low-quality images. We make several interesting findings that provide insight into existing problems with FH models and point to future research directions.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
