2020 • Wearable • Electronics
Eye Spy is a wearable employing a StyleGAN-based neural network trained on a hybrid dataset of moth eyespots and human ocular anatomy. The system generates continuous streams of synthetic, biomimetic imagery displayed across five embedded micro-LCD panels arranged to orbit the wearer’s face. These evolving “generatmoths” exploit vulnerabilities in computer-vision algorithms by presenting organic, eye-like forms that register as gaze patterns without providing consistent facial landmarks required for recognition. Each screen operates on an independent seed, producing non-repeating visual outputs that evolve in real time through latent-space interpolation. This continuous variation prevents pattern-tracking systems from stabilizing on a fixed identity, effectively creating a dynamic optical shield. The device’s onboard edge processor coordinates imagery updates and synchronizes display brightness to ambient lighting conditions, ensuring optimal visibility to machine vision while remaining subtle to the human observer. 
