Ziyun Wang

Ziyun Wang

Auto-generated profile, based on publicly available information

Ziyun Wang

University of Pennsylvania,Department of Computer and Information Science,Philadelphia,PA,19104

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

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

Representative Publications

View All Publications
23 May 2021
Verified
Conference Paper
Open access
AZAlex Zihao Zhu
Alex Zihao Zhu
ZWZiyun Wang
Ziyun Wang
KKKaung Khant
Kaung Khant
KDKostas Daniilidis
Kostas Daniilidis
Event cameras provide a number of benefits over traditional cameras, such as the ability to track incredibly fast motions, high dynamic range, and low power consumption. However, their application into computer vision problems, many of which are primarily dominated by deep learning solutions, has been limited by the lack of labeled training data for events. In this work, we propose a method which leverages the existing labeled data for images by simulating events from a pair of temporal image frames, using a convolutional neural network. We train this network on pairs of images and events, using an adversarial discriminator loss and a pair of cycle consistency losses. The cycle consistency losses utilize a pair of pre-trained self-supervised networks which perform optical flow estimation and image reconstruction from events, and constrain our network to generate events which result in accurate outputs from both of these networks. Trained fully end to end, our network learns a generative model for events from images without the need for accurate modeling of the motion in the scene, exhibited by modeling based methods, while also implicitly modeling event noise. Using this simulator, we train a pair of downstream networks on object detection and 2D human pose estimation from events, using simulated data from large scale image datasets, and demonstrate the networks' abilities to generalize to datasets with real events. The code and dataset in this paper are available here: https://github.com/alexzzhu/EventGAN.

2 February 2021 in Journal of Lightwave Technology
Verified
Journal Article
Closed access
SZShiyuan Zhao
Shiyuan Zhao
JCJiwen Cui
Jiwen Cui
ZWZhanjun Wu
Zhanjun Wu
ZWZiyun Wang
Ziyun Wang
JTJiubin Tan
Jiubin Tan
Fiber deformation measurement is crucial in the structure health monitor, which reflects the stress, strain or deformation of the structure where the fiber is laid. The principle of the fiber deformation measurement by the optical phase tracking in OFDR scheme is described and the data processing algorithm is demonstrated in details to receive accurate results. The differential phase is acquired by making a difference between the phase spectrum of the reference and measurement states. A robust phase unwrapping method based on the density distribution is proposed to get the whole unwrapped low-noise phase of the fiber. Additionally, the phase decoherence caused by the scatter unit misalignment induced by the fiber deformation is analyzed and the phase correction method is proposed to improve the phase coherence. The fiber deformation can be tracked along the fiber with a resolution of tens of microns, depending on the set parameters of the OFDR system. Moreover, the strain distribution is also derived from the differential phase. In the experiments, the measurements of the standard single-mode fiber with uniform and non-uniform deformation have been accomplished in a spatial resolution of 40 μm. The noise is about 0.1 μm in a stretching range of about 40 μm. Simultaneously, hundreds of microstrain are obtained and show a high accuracy with a resolution of the level of millimeter.