I'm a Ph.D. student at Carnegie Mellon University, advised by
Prof. Aswin C. Sankaranarayanan.
My thesis has been focused on pushing the frontier of mobile photography,
by jointly design camera optics, sensor, and efficient deep restoration algorithms.
I have won best paper award at ICCP in 2021
and Google Anita Borg Scholarship in 2016.
My research interest includes:
novel cameras design, such as under-display cameras
efficient deep learning for camera ISP and downstream tasks
extend the capability of photography using diffusion models
Prior to that, I obtained my M.S. degree from the Robotics Institute at CMU and my
M.S. thesis
was on 3D object detection, segmentation, and classfication for dense 3D volumes.
I obtained my B.Eng. degree from School of Software Engineering at Tongji University, working with
Prof. Lin Zhang.
Under-display cameras is an emerging type of camera that can image a scene through the openings of the display screens.
This work proposes a novel design of diffractive optical elements that effectively renders the display screen invisible
from the camera's perspective and keeps the display quality high at the same time.
Designing Display Pixel Layouts for Under-Panel Cameras Anqi Yang,
Aswin C. Sankaranarayanan
TPAMI / ICCP, 2021   (Best Paper Award) paper
/
video
/
pressKey words: Joint optimization of camera and image restoration algorithms, camera ISP, deep learning-based deblurring, hardware prototype, wave optics
The opening patterns on the display plays a crucial role in determining diffractive blur in under-panel cameras.
This work proposes a suite of modifications to the display opening patterns that significantly increases the
imaging quality of UDCs.
This work addresses the challenging task of automated object detection in aviation screening with 3D computed tomography scans.
We propose a slice-and-fuse framework, that slices 3D volumes into 2D slices, leverages efficient 2D detectors (e.g. RetinaNet),
and fuse 2D predictions back to 3D.
Pain Expression Recognition using Occluded Face
Ahmed Ashraf,
Anqi Yang,
Babak Taati
FG, 2019   (Best Poster Award) paperKey words: Facial landmarks, biometrics, machine learning for regression and classification
This paper presents a preliminary study on automatic recognition of facial expression
in ICU settings, where the patient face is partially occluded by surgery masks.
This paper presents a novel contacless palmprint acquisition system, a large-scale palmprint benchmark,
as well as a novel machine learning-based palmprint classification algorithm.
Awards
ICCP Best Paper Award, 2021
Liang Ji-Dian Fellowship, 2022
Tan Endowed Graduate Fellowship, 2022
Google Anita Borg Memorial Scholarship, 2016
Miscellanea
Recitation TA, CMU-18290 Signals and Systems
Recitation TA, CMU-18793 Image and Video Processing
Reviewer for TPAMI, TIP, CVPR, ICCV, ECCV, BMVC, WACV