In‐Sensor Reservoir Computing Using Ferroelectric Optoelectronic Synapse for Near‐Infrared Face Recognition
Wenyu Songlu, Quan Chen, Yi Heng Cao, Han Wang, Han Chen, Weihui Sang, Yang Gan, Shijia Tian, Yuan Yu, Xin Zhou, Yukang Chen, Shuming Guo, Lin Wang, Xumeng Zhang, Tao Liu, Du Xiang · Advanced Intelligent Systems · 2025
Face recognition in the infrared band plays a vital role in diverse biometric technologies ranging from authentication security to daily healthcare. Current infrared recognition systems are built on the von Neumann architecture and complementary metal‐oxide‐semiconductor image sensors, which encounter limited spectral adaptability with serious data transmission delay and high power consumption. Here, a compact in‐sensor reservoir computing (RC) system is demonstrated using van der Waals α ‐In 2 Se 3 optoelectronic device to implement the face recognition task in near‐infrared band. Due to the light‐induced temporal ferroelectric polarization field in α ‐In 2 Se 3 , the nonlinear photoresponse and short‐term memory characteristics are achieved in the device with low optical energy consumption of ≈15 pJ. Additionally, the device exhibits diverse synaptic plasticity with excellent separation of the optical information based on the dual‐feature strategy. An in‐sensor RC system is constructed through employing the α ‐In 2 Se 3 optoelectronic synapse as the physical node, which achieves 99.5% recognition accuracy with fast training speed. The robustness of the system is further verified by the preservation of accuracy at 87.1% under high noise level of 20%. This work envisions ferroelectric optoelectronic synapse as potential building blocks for in‐sensor RC system, paving the way for achieving high‐accuracy recognition system in extended spectral range.