Deep visual features are increasingly used as the interface in vision systems, motivating the need to describe feature characteristics and control feature quality for machine perception. Just noticeable difference (JND) characterizes the maximum imperceptible distortion for images under human or machine vision. Extending it to deep visual features naturally meets the above demand by providing a task-aligned tolerance boundary in feature space, offering a practical reference for controlling feature quality under constrained resources. We propose FeatJND, a task-aligned JND formulation that predicts the maximum tolerable per-feature perturbation map while preserving downstream task performance. We propose a FeatJND estimator at standardized split points and validate it across image classification, detection, and instance segmentation. Under matched distortion strength, FeatJND-based distortions consistently preserve higher task performance than unstructured Gaussian perturbations, and attribution visualizations suggest FeatJND can suppress non-critical feature regions. As an application, we further apply FeatJND to token-wise dynamic quantization and show that FeatJND-guided step-size allocation yields clear gains over random step-size permutation and global uniform step size under the same noise budget. Our code will be released after publication.

翻译：深度视觉特征日益成为视觉系统中的关键接口，这促使我们需要描述特征特性并控制面向机器感知的特征质量。恰可察觉差异（JND）描述了人类或机器视觉下图像的最大不可感知失真。将其扩展至深度视觉特征，通过在特征空间中提供任务对齐的容差边界，自然满足了上述需求，为资源受限条件下的特征质量控制提供了实用参考。我们提出FeatJND，这是一种任务对齐的JND建模方法，能够预测在保持下游任务性能前提下的最大可容忍逐特征扰动图。我们设计了标准化分割点处的FeatJND估计器，并在图像分类、检测和实例分割任务上进行了验证。在相同失真强度下，基于FeatJND的失真相较于非结构化高斯扰动能持续保持更高的任务性能，归因可视化表明FeatJND能够抑制非关键特征区域。作为应用案例，我们进一步将FeatJND应用于面向token的动态量化，结果表明在相同噪声预算下，FeatJND引导的步长分配策略相较于随机步长置换和全局均匀步长方案均能取得显著增益。我们的代码将在论文发表后开源。