Vitaly Ablavsky

Timeline

1. 1992–1996

   Mathematics and computer science

   He earns a BA in mathematics at Brandeis University in 1992 and an MS in computer science at the University of Massachusetts Amherst in 1996.

2. 2011

   A doctorate at Boston University

   Before his doctorate, he contributes to machine vision research and development at Charles River Analytics, Cognex and Amerinex Applied Imaging with ADCIS. In 2011, he earns a computer science PhD in Boston University's Image and Video Computing group, advised by Stan Sclaroff, studying graphical models for tracking. He then holds a postdoctoral appointment in EPFL's CVLab with Pascal Fua and Vincent Lepetit, working on activity recognition and transfer learning.

3. 2014

   Tracking balls through team play

   In 2014, he coauthors “Take your Eyes off the Ball: Improving Ball-Tracking by Focusing on Team Play” with Xinchao Wang, Horesh Ben Shitrit and Pascal Fua. The method tracks the ball by deciding which player has possession, and is tested on basketball and soccer videos.

4. 2015

   Research at STR

   In 2015, he works at Systems & Technology Research (STR) and serves on the technical program committee of the Fusion 2015 conference. As a senior member of technical staff in STR's Video and Image Understanding group, he contributes to research and development and leads technical areas on DARPA and IARPA programs.

5. August 2019

   Recycling research at Boston University

   In August 2019, he is a senior research scientist in computer science at Boston University and an investigator on an NSF-funded study of human–robot collaboration in recycling. The project examines whether robots can improve waste sorting, recycling quality and working conditions.

6. 2021

   Siamese Natural Language Tracker

   In 2021, he coauthors the Siamese Natural Language Tracker with Qi Feng, Qinxun Bai and Stan Sclaroff at CVPR. The system combines visual information and natural language descriptions to track objects, and runs at 50 frames per second on a single GPU in the reported tests.

7. 2022

   The ZeroWaste dataset

   In 2022, he coauthors the ZeroWaste dataset paper at CVPR. ZeroWaste provides images from an industrial recycling setting for detecting and segmenting waste, including deformable and translucent objects in cluttered scenes.

8. 2024

   Learning to track with SSP-GNN

   In 2024, he coauthors “SSP-GNN: Learning to Track via Bilevel Optimization” with Griffin Golias and Masa Nakura-Fan. The method combines a shortest-path tracking algorithm with a graph neural network whose parameters are learned from example tracks and detections.

9. April 2026

   SAGA-ReID for person recognition

   In April 2026, he coauthors a paper introducing SAGA-ReID with Aotian Zheng, Winston Sun, Bahaa Alattar and Jenq-Neng Hwang. The method uses CLIP features to recognize the same person across images, with experiments that test cases where parts of the person are hidden.

10. As of October 2026

    Research roles at Washington

    As of October 2026, he serves as a principal research scientist in the University of Washington's Applied Physics Laboratory, with an affiliation in Environmental & Information Systems. He is also an affiliate assistant professor in Electrical and Computer Engineering.