Deep learning-developed multi-light source discrimination capability of stretchable capacitive photodetector
Su Bin Choi, Jun Sang Choi, Hyun Sik Shin, Jeong-Won Yoon, Youngmin Kim, Jong‐Woong Kim
Abstract
We introduce a novel stretchable photodetector with enhanced multi-light source detection, capable of discriminating light sources using artificial intelligence (AI). These features highlight the application potential of deep learning enhanced photodetectors in applications that require accurate for visual light communication (VLC). Experimental results showcased its excellent potential in real-world traffic system. This photodetector, fabricated using a composite structure of silver nanowires (AgNWs)/zinc sulfide (ZnS)-polyurethane acrylate (PUA)/AgNWs, maintained stable performance under 25% tensile strain and 2 mm bending radius. It shows high sensitivity at both 448 and 505 nm wavelengths, detecting light sources under mechanical deformations, different wavelengths and frequencies. By integrating a one-dimensional convolutional neural network (1D-CNN) model, we classified the light source power level with 96.52% accuracy even the light of two wavelengths is mixed. The model’s performance remains consistent across flat, bent, and stretched states, setting a precedent for flexible electronics combined with AI in dynamic environments.