SOURCE-LINKED INTELLIGENCE

Feature-Spectral Fragility in Segmentation: Dataset Dependence, Architecture-Specific Localization, and Spectral Correlates

arXiv · AI, language, vision and robotics · article · Aug 29, 2026 · UTC

Robustness of segmentation models is commonly assessed through input-domain perturbations, while dependence on frequency content within learned feature representations remains less understood. We probe this dependence using targeted post-training low-pass interventions on internal representations of three segmentation architectures, ResNet50-UNet (CNN), VM-UNet (SSM), and Swin-UNETR (Transformer), across CVC-ClinicDB and ISIC2018, with headline evaluations performed on untouched held-out test sets. At cutoff rho=0.25, feature-domain low-pass filtering causes severe degradation on CVC: Dice dro

Read original source ↗ Open in workspace

recordType

paper

region

Global