Bio-inspired optoelectronic devices and systems for energy-efficient in-sensor computing
Xiaoting Wang, Heyi Huang, Jianshi Tang, Ruofei Hu, Yu Du, Yuyan Wang, Bin Gao, Qian He, Huaqiang Wu
Abstract
AI-driven machine vision faces critical energy and latency challenges from exponential growth in visual data. This review proposes bio-inspired energy-efficient in-sensor computing utilizing emerging optoelectronic memristors, examining neural network architectures (fully connected/convolutional, recurrent, and spiking neural networks) for static, motion, and event-driven processing that directly emulates biological visual pathways. Critical integration challenges and strategic roadmaps are systematically analysed to achieve cortex-level energy efficiency in next-generation vision systems.
Topics & Concepts
OptoelectronicsComputer scienceMaterials scienceEngineering physicsPhysicsAdvanced Memory and Neural ComputingNeural Networks and Reservoir ComputingPhotoreceptor and optogenetics research