Owl-vision-inspired near sensor computing

Zishen Zhao; Yixin Cao; Shuaiwei Huang; Kaiwen Fan; Yuyao Ding; Ganggui Zhu; Linhui Li; Qing Liu; Wei Deng; Chun Zhao; Mario Lanza

doi:10.1038/s41467-026-69123-7

Owl-vision-inspired near sensor computing

Nat Commun. 2026 Feb 12. doi: 10.1038/s41467-026-69123-7. Online ahead of print.

Authors

Zishen Zhao^#^{1

2}, Yixin Cao^#^{1

3}, Shuaiwei Huang¹, Kaiwen Fan¹, Yuyao Ding¹, Ganggui Zhu³, Linhui Li³, Qing Liu¹, Wei Deng⁴, Chun Zhao⁵, Mario Lanza^{6

7}

Affiliations

¹ School of Advanced Technology, Xi'an Jiaotong-Liverpool University, Suzhou, China.
² Department of Electrical Engineering and Electronics, University of Liverpool, Liverpool, UK.
³ Department of Military Biomedical Engineering, Fourth Military Medical University, Xi'an, China.
⁴ Institute of Functional Nano & Soft Materials (FUNSOM), Jiangsu Key Laboratory for Carbon-Based Functional Materials & Devices, Soochow University, Suzhou, Jiangsu, China. dengwei@suda.edu.cn.
⁵ School of Advanced Technology, Xi'an Jiaotong-Liverpool University, Suzhou, China. chun.zhao@xjtlu.edu.cn.
⁶ Department of Materials Science and Engineering, National University of Singapore, Singapore, Singapore. mlanza@nus.edu.sg.
⁷ Institute for Functional Intelligent Materials, National University of Singapore, Singapore, Singapore. mlanza@nus.edu.sg.

^# Contributed equally.

PMID: 41680170
DOI: 10.1038/s41467-026-69123-7

Abstract

Passive target detection in photon-starved environments is crucial to expand machine vision in a wide range of applications such as precision guidance, intelligent surveillance, and early warning. Here, inspired by owl vision, we report a bimodal synaptic transistor with an optoelectronic decoupling mechanism, enabling parallel photonic perception and electrical plasticity emulation. As a result, the device exhibits a high active adaptation index of approximately 331, alongside the ability to perceive low light intensities as faint as 0.146 nW cm^-2. We also achieve cyclically stable synaptic weight modulation with long-term enhancement and inhibition, and verify the feasibility of weight deployment across three basic artificial neural levels over a light intensity range of 0.146-11.70 nW cm^-2, via adaptive contrast enhancement. The owl-vision-inspired device establishes a hardware foundation forward in energy-efficient and low-light image processing for neuromorphic vision sensors.