Shih-Yao Lin

Shih-Yao Lin

Auto-generated profile, based on publicly available information

Shih-Yao Lin

Department of Physics, National Taiwan University, Taipei 10617, Taiwan

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 August 2018 in ACS Applied Nano Materials
Verified
Journal Article
Closed access
SLShih-Yao Lin
Shih-Yao Lin
Golam HaiderGolam Haider
Golam Haider
YLYu-Ming Liao
Yu-Ming Liao
CCCheng-Han Chang
Cheng-Han Chang
WLWei-Ju Lin
Wei-Ju Lin
YCYang-Fang Chen
Yang-Fang Chen

7 August 2019 in ACS Nano
Verified
Journal Article
Closed access
YHYun-Tzu Hsu
Yun-Tzu Hsu
CTChia-Tse Tai
Chia-Tse Tai
HWHsing-Mei Wu
Hsing-Mei Wu
CHCheng-Fu Hou
Cheng-Fu Hou
YLYu-Ming Liao
Yu-Ming Liao
YCYang-Fang Chen
Yang-Fang Chen

12 October 2020
Verified
Conference Paper
Open access
ZWZhenyu Wu
Zhenyu Wu
DHDuc Hoang
Duc Hoang
SLShih-Yao Lin
Shih-Yao Lin
YXYusheng Xie
Yusheng Xie
CJChen Liang Jian
Chen Liang Jian
WFWei Fan
Wei Fan
Estimating the 3D hand pose from a monocular RGB image is important but challenging. A solution is training on large-scale RGB hand images with accurate 3D hand keypoint annotations. However, it is too expensive in practice. Instead, we develop a learning-based approach to synthesize realistic, diverse, and 3D pose-preserving hand images under the guidance of 3D pose information. We propose a 3D-aware multi-modal guided hand generative network (MM-Hand), together with a novel geometry-based curriculum learning strategy. Our extensive experimental results demonstrate that the 3D-annotated images generated by MM-Hand qualitatively and quantitatively outperform existing options. Moreover, the augmented data can consistently improve the quantitative performance of the state-of-the-art 3D hand pose estimators on two benchmark datasets. The code will be available at https://github.com/ScottHoang/mm-hand.

21 July 2020 in ACS Omega
Verified
Journal Article
Open access
HHHan-Wen Hu
Han-Wen Hu
Golam HaiderGolam Haider
Golam Haider
YLYu-Ming Liao
Yu-Ming Liao
PRPradip Kumar Roy
Pradip Kumar Roy
HLHung-I. Lin
Hung-I. Lin
YCYang-Fang Chen
Yang-Fang Chen
Total internal reflection is one of the most important phenomena when a propagated wave strikes a medium boundary, which possesses a wide range of applications spanning from optical communication to a fluorescence microscope. It has also been widely used to demonstrate conventional laser actions with resonant cavities. Recently, cavity-free stimulated emission of radiation has attracted great attention in disordered media because of several exciting physical phenomena, ranging from Anderson localization of light to speckle-free imaging. However, unlike conventional laser systems, the total internal reflection has never been implemented in the study of laser actions derived from randomly distributed media. Herein, we demonstrate an ultra-low threshold cavity-free laser system using air bubbles as scattering centers in which the total internal reflection from the surface of air bubbles can greatly reduce the leakage of the scattered beam energy and then enhance light amplification within a coherent closed loop. Our approach provides an excellent alternative for the manipulation of optical energy flow to achieve ultra-low threshold cavity-free laser systems, which should be very useful for the development of high performance optoelectronic devices.

9 December 2021 in arXiv
Verified
Preprint
Open access
PDF file
HMHaoyu Ma
Haoyu Ma
CJChen Liang Jian
Chen Liang Jian
DKDeying Kong
Deying Kong
ZWZhe Wang
Zhe Wang
XLXingwei Liu
Xingwei Liu
XXXiaohui Xie
Xiaohui Xie
Estimating the 2D human poses in each view is typically the first step in calibrated multi-view 3D pose estimation. But the performance of 2D pose detectors suffers from challenging situations such as occlusions and oblique viewing angles. To address these challenges, previous works derive point-to-point correspondences between different views from epipolar geometry and utilize the correspondences to merge prediction heatmaps or feature representations. Instead of post-prediction merge/calibration, here we introduce a transformer framework for multi-view 3D pose estimation, aiming at directly improving individual 2D predictors by integrating information from different views. Inspired by previous multi-modal transformers, we design a unified transformer architecture, named TransFusion, to fuse cues from both current views and neighboring views. Moreover, we propose the concept of epipolar field to encode 3D positional information into the transformer model. The 3D position encoding guided by the epipolar field provides an efficient way of encoding correspondences between pixels of different views. Experiments on Human 3.6M and Ski-Pose show that our method is more efficient and has consistent improvements compared to other fusion methods. Specifically, we achieve 25.8 mm MPJPE on Human 3.6M with only 5M parameters on 256 x 256 resolution.