(Peer-Reviewed) High performance "non-local" generic face reconstruction model using the lightweight Speckle-Transformer (SpT) UNet
Yangyundou Wang 王杨云逗 ¹ ², Hao Wang 王昊 ³, Min Gu 顾敏 ¹ ²
¹ Institute of Photonic Chips, University of Shanghai for Science and Technology, Shanghai 200093, China
中国 上海 上海理工大学光子芯片研究院
² Centre for Artificial-Intelligence Nanophotonics, School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
中国 上海 上海理工大学 光电信息与计算机工程学院 人工智能纳米光子学中心
³ School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
中国 上海 上海理工大学 光电信息与计算机工程学院
Opto-Electronic Advances, 2023-02-25
Abstract
Significant progress has been made in computational imaging (CI), in which deep convolutional neural networks (CNNs) have demonstrated that sparse speckle patterns can be reconstructed. However, due to the limited “local” kernel size of the convolutional operator, for the spatially dense patterns, such as the generic face images, the performance of CNNs is limited.
Here, we propose a “non-local” model, termed the Speckle-Transformer (SpT) UNet, for speckle feature extraction of generic face images. It is worth noting that the lightweight SpT UNet reveals a high efficiency and strong comparative performance with Pearson Correlation Coefficient (PCC), and structural similarity measure (SSIM) exceeding 0.989, and 0.950, respectively.
A 4096-element 3D-integrated Si-SiN optical phased array for high-power coherent LiDAR
Han Wang, Weimin Xie, Xin Yan, Jiaqi Li, Youxi Lu, Ping Jiang, Feng Li, Kai Jin, Xu Yang, Jiali Jiang, Keran Deng, Weishuai Chen, Jing Luo, Li Jin, Junbo Feng, Kai Wei
Opto-Electronic Technology
2026-03-20
High-speed and large-capacity visible light communication for 6G: advances and perspectives
Nan Chi, Zhilan Lu, Fujie Li, Haoyu Zhang, Yunkai Wang, Xinyi Liu, Zhiwu Chen, Zhe Feng, Zhuoran Hu, Zhixue He, Ziwei Li, Chao Shen, Junwen Zhang
Opto-Electronic Technology
2026-03-20
Holotomography-driven learning unlocks in-silico staining of single cells in flow cytometry by avoiding fluorescence co-registration
Daniele Pirone, Giusy Giugliano, Michela Schiavo, Annalaura Montella, Martina Mugnano, Vincenza Cerbone, Maddalena Raia, Giulia Scalia Ivana Kurelac, Diego Luis Medina, Lisa Miccio Mario Capasso, Achille Iolascon, Pasquale Memmolo, Pietro Ferraro
Opto-Electronic Science
2026-02-25
