MMM
YYYY
Integrated photonic convolution acceleration core for wearable devices
用于可穿戴设备的集成光子卷积加速核心
ウェアラブルデバイス用の統合フォトニック畳み込み加速コア
웨어러블 장치용 통합 광자 볼륨 가속 코어
Núcleo de aceleración de convolución fotónica integrado para dispositivos portátiles
Noyau d'accélération de convolution photonique intégré pour wearables
Интегрированное ядро ускорения свертывания фотонов для носимых устройств
Baiheng Zhao 赵柏衡 ¹, Junwei Cheng 成骏伟 ¹, Bo Wu 吴波 ¹, Dingshan Gao 郜定山 ¹, Hailong Zhou 周海龙 ¹, Jianji Dong 董建绩 ¹ ²
¹ Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wuhan 430074, China
中国 武汉 华中科技大学 武汉光电国家实验室
² Optics Valley Laboratory, Wuhan 430074, China
中国 武汉 湖北光谷实验室
Opto-Electronic Science, 28 November 2023
Abstract

With the advancement of deep learning and neural networks, the computational demands for applications in wearable devices have grown exponentially. However, wearable devices also have strict requirements for long battery life, low power consumption, and compact size.

In this work, we propose a scalable optoelectronic computing system based on an integrated optical convolution acceleration core. This system enables high-precision computation at the speed of light, achieving 7-bit accuracy while maintaining extremely low power consumption. It also demonstrates peak throughput of 3.2 TOPS (tera operations per second) in parallel processing.

We have successfully demonstrated image convolution and the typical application of an interactive first-person perspective gesture recognition application based on depth information. The system achieves a comparable recognition accuracy to traditional electronic computation in all blind tests.
Opto-Electronic Science_1
Opto-Electronic Science_2
Opto-Electronic Science_3
Opto-Electronic Science_4
Reviews and Discussions
https://www.hotpaper.io/index.html
Dynamic interactive bitwise meta-holography with ultra-high computational and display frame rates
Multi-dimensional multiplexing optical secret sharing framework with cascaded liquid crystal holograms
Physics-informed deep learning for fringe pattern analysis
Advancing computer-generated holographic display thanks to diffraction model-driven deep nets
Wide-spectrum optical synthetic aperture imaging via spatial intensity interferometry
Deep learning enabled single-shot absolute phase recovery in high-speed composite fringe pattern profilometry of separated objects
Exceptional-point-enhanced sensing in an all-fiber bending sensor
All-optical object identification and three-dimensional reconstruction based on optical computing metasurface
Ferroelectrically modulate the Fermi level of graphene oxide to enhance SERS response
Photonic integrated neuro-synaptic core for convolutional spiking neural network
Paper-based WS₂ photodetectors fabricated by all-dry techniques
Spatiotemporal hemodynamic monitoring via configurable skin-like microfiber Bragg grating group



Previous Article                                Next Article
About
|
Contact
|
Copyright © Hot Paper