Foveated Retinotopy Improves Classification and Localization in CNNs

A foveated retinotopic transform applied to a ResNet input preserves classification accuracy and improves robustness to scale and rotation, while enabling implicit object localization. getnews.me/foveated-retinotopy-impr... #foveatedvision #cnn

Foveated Vision Enhances Self‑Supervised Object Learning in New Study

Simulated foveated vision improves self‑supervised object learning on egocentric video, the study reports. Published Sep 2025, accepted at IEEE ICDL 2025. Read more: getnews.me/foveated-vision-enhances... #foveatedvision #selfsupervised #icdl