通过可微光线传输实现场景参数显著性
Scene Parameter Saliency via Differentiable Light Transport
查看机构详情
- Simon Fraser University（西蒙弗雷泽大学）
机构由 AI 辅助整理，请以论文原文为准。
浏览论文内容
中文总结 AI 辅助
研究通过可微光线传输实现场景参数显著性，计算针对不同目标的度量显著性图，发现同一场景不同度量下显著性排名差异大，显著性图因度量而异，结果表明可微渲染器的导数图像对场景理解有重要价值。
中文摘要 AI 辅助
基于梯度的显著性方法揭示哪些输入特征对神经网络输出影响最大，是模型可解释性的标准工具。我们发现，常用于参数优化的可微渲染器会产生类似的显著性形式：给定在渲染图像上评估的任何标量度量，单次反向模式微分传递会产生每个参数的梯度，以识别哪些场景元素对该度量影响最大。我们将这些梯度场称为度量显著性图。与通过学习权重传播归因的神经显著性不同，度量显著性通过图像形成过程本身传播，包括多次反射光线传输，捕捉到难以通过人工检查发现的参数依赖性。我们针对不同性质的目标计算度量显著性图：心理视觉眩光指数、平均场景亮度和神经感知分数。对于同一场景，不同度量的显著性排名差异很大，对一个目标起主导作用的参数对另一个目标可忽略不计。显著性图特定于度量，而非场景的固有属性。我们的结果表明，可微渲染器产生的导数图像对于场景理解与它们设计生成的原始图像一样具有信息价值。
英文摘要
Gradient-based saliency methods reveal which input features most influence a neural network's output, and are a standard tool for model interpretability. We observe that differentiable renderers, which are conventionally used for parameter optimisation, produce an analogous form of saliency: given any scalar metric evaluated on a rendered image, a single reverse-mode differentiation pass yields per-parameter gradients that identify which scene elements most influence the metric. We call these gradient fields metric saliency maps. Unlike neural saliency, which propagates attribution through learned weights, metric saliency propagates through the image formation process itself, including multi-bounce light transport, capturing parameter dependencies that are semi-opaque to manual inspection. We compute metric saliency maps for qualitatively different objectives: psychovisual glare indices, mean scene luminance, and neural perceptual scores. The saliency rankings differ substantially across metrics for the same scene, with parameters that dominate one objective being negligible for another. The saliency map is specific to the metric, not an intrinsic property of the scene. Our results suggest that differentiable renderers produce derivative images that are as informative for scene understanding as the primal images they were designed to generate.
