CSFlow: 将流匹配与人类对比敏感度对齐
CSFlow: Aligning Flow Matching with Human Contrast Sensitivity
AI总结：
提出CSFlow加权方案，通过将人类对比敏感度函数与流匹配的迭代去噪步骤对齐，在傅里叶空间中引入软自回归结构，提升生成图像的视觉真实感，FID降低4.7%，Inception Score提升2.2%。
摘要：
我们引入了对比敏感流（CSFlow），这是一种将人眼的对比敏感度函数（CSF）与流匹配的迭代去噪步骤联系起来的加权方案。由于真实世界图像将信号集中在低空间频率，这些分量在连续扩散过程中比高频分量更早达到高信噪比。当使用扩散或流匹配模型生成图像时，这会在傅里叶空间中诱导一种软自回归结构，其中粗略的图像内容在精细细节之前稳定。同时，人类视觉系统对空间频率的敏感度不均：极低和极高的频率需要显著更高的对比度才能被感知。我们首次通过两个贡献将这些观察结果融合在一起：（1）一个估计每个反向流区间生成哪些频率的度量，以及（2）通过将每个噪声级别生成的频率与人类对比敏感度对齐获得的时间步权重。我们通过实验验证了我们的贡献，表明这些权重可以通过仅推理时间步修改或短时微调，将FID降低4.7%，Inception Score提高2.2%，GenEval分数提高2.5%，从而改善生成性能。定性上，我们发现我们的CSFlow权重导致生成的图像具有更好的视觉真实感和更少的卡通外观。
英文摘要：
We introduce Contrast Sensitive Flow (CSFlow), a weighting scheme that connects the human eye's Contrast Sensitivity Function (CSF) to the iterative denoising steps of flow matching. Because real-world images concentrate signal at low spatial frequencies, these components reach high signal-to-noise ratio earlier during continuous diffusion than high-frequency components. When generating images with diffusion or flow matching models, this induces a soft autoregressive structure in Fourier space, where coarse image content stabilizes before fine detail. Meanwhile, the human visual system is unequally sensitive to spatial frequencies: very low and very high frequencies require significantly higher contrast to be perceived. We for the first time merge these observations through two contributions: (1) a metric that estimates which frequencies are generated at each reverse flow interval and (2) timestep weights obtained by aligning the frequencies generated at each noise level with human contrast sensitivity. We validate our contributions experimentally showing that these weights can improve generative performance by lowering FID by 4.7%, increasing Inception Score by 2.2% and improving GenEval scores by 2.5% using inference-only timestep modification or short fine-tuning. Qualitatively, we find that our CSFlow weights lead to better visual realism and less cartoonish appearance of generated images.
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。
