About Me

I am currently looking for a job starting September 2021 or later. If you have an opening, please contact me at szetor@umich.edu!

I am a Computer Science and Engineering Ph.D. candidate at the University of Michigan in Ann Arbor. I am advised by Prof. Jason J. Corso and Prof. Honglak Lee. My current research explores conditional video generation as applied to tasks like video prediction, inpainting, and style transfer.

In 2019, I did an internship at Samsung Semiconductor, Inc. in San Diego, CA. Advised by Dr. Mostafa El-Khamy, I developed a method to apply image style transfer and inpainting models to videos in a temporally-consistent manner.

In 2017, I did an internship at Toyota Research Institute in Cambridge, MA under the advisement of Dr. Simon Stent and Dr. German Ros, where I compared the modeling and generalization performance of multiple state-of-the-art video prediction networks with a novel dataset.

In 2015, I graduated summa cum laude from the University of Massachusetts in Amherst, where I received B.S. degrees in Computer Science and Mathematics. At UMass, I did research as part of the RIPPLES lab under Prof. W. Richards Adrion and Prof. Paul E. Dickson. I worked on various aspects of the lab's Presentations Automatically Organized from Lectures (PAOL) system, including video conversion, whiteboard processing, multithreading, and the graphical user interface. For my Honors Thesis project, I proposed and evaluated a novel technique for segmenting whiteboard marker strokes in real time. Additionally, I was briefly a member of the Center for e-Design under the instruction of Prof. Jack C. Wileden and Prof. Sundar Krishnamurty, where I worked on a program that converted models between Computer-Aided Design systems.

Selected Publications

PDF HyperCon: Image-To-Video Model Transfer for Video-To-Video Translation Tasks
Ryan Szeto, Mostafa El-Khamy, Jungwon Lee, and Jason J. Corso
IEEE Winter Conference on Applications of Computer Vision, 2021
[ BibTeX ]
LINK A Temporally-Aware Interpolation Network for Video Frame Inpainting
Ryan Szeto, Ximeng Sun, Kunyi Lu, and Jason J. Corso
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020
[ BibTeX ]
PDF A Dataset To Evaluate The Representations Learned By Video Prediction Models
Ryan Szeto, Simon Stent, German Ros, and Jason J. Corso
International Conference on Learning Representations (Workshop Track), 2018
[ BibTeX ]
PDF Click Here: Human-Localized Keypoints as Guidance for Viewpoint Estimation
Ryan Szeto and Jason J. Corso
IEEE International Conference on Computer Vision, 2017
[ BibTeX ]

Awards and Honors

  • NSF Graduate Research Fellowship 2017 - Honorable Mention (UMich)
  • Outstanding Achievement in Artificial Intelligence Award (UMass)
  • Honors Dean's Award (UMass)
  • Honors Research Grant (UMass)
  • Research Assistant Fellowship (UMass)

News

04/07/2020

The extended version of “A Temporally-Aware Interpolation Network for Video Frame Inpainting” will appear in the May 2020 issue of IEEE Transactions on Pattern Analysis and Machine Intelligence!

03/26/2019

I will be doing an internship with Samsung Semiconductor Inc. in San Diego, CA this summer!

09/22/2018

The paper “A Temporally-Aware Interpolation Network for Video Frame Inpainting”, written by me and my equal co-author Ximeng Sun, has been accepted to ACCV 2018!