Abstract Summary/Description
We introduce a screen-to-camera communication system that embeds data in real-world settings by leveraging temporal flicker fusion within the OKLAB color space. Through spatially adaptive flickering and encoding information in distinct pixel region shapes, Revelio achieves data embedding that is visually imperceptible yet highly resilient to noise, asynchronicity, and distortions common in screen-camera channels, allowing for reliable decoding by standard smartphone cameras. Powered by a two-stage neural network, our decoding process employs a weighted differential accumulator to enhance frame detection and symbol recognition accuracy. Early experiments showcase the potential of interactive television, providing a seamless method for transmitting meta-information without interrupting the viewing experience.
