APL-UW Home

Jobs
About
Campus Map
Contact
Privacy
Intranet

Vitaly Ablavsky

Principal Research Scientist

Email

vablavsky@apl.washington.edu

Phone

206-616-0380

Research Interests

Dr. Ablavsky recently joined APL-UW. Previously, he was a senior research scientist in the Image and Video Computing (IVC) group of the Department of Computer Science at Boston University.

Ablavsky's research has focused on machine learning, computer vision, and autonomous systems. His broader interests include the application of artificial intelligence to problems in diverse domains and the role of AI in our society.

Education

B.A. Mathematics, Brandeis University, 1992

M.S. Computer Science, University of Massachusetts at Amherst, 1996

Ph.D. Computer Science, Boston University, 2011

Vitaly Ablavsky's Website

https://www.corvidim.net/ablavsky/

Publications

2000-present and while at APL-UW

Learning to separate: Detecting heavily-occluded objects in urban scenes

Yang, C., V. Ablavsky, K. Wang, Q. Feng, and M. Betke, "Learning to separate: Detecting heavily-occluded objects in urban scenes," in Proc., 16 European Conference on Computer Vision, 23-28 August (Springer, 2020).

More Info

23 Aug 2020

While visual object detection with deep learning has received much attention in the past decade, cases when heavy intra-class occlusions occur have not been studied thoroughly. In this work, we propose a novel Non-Maximum-Suppression (NMS) algorithm that dramatically improves the detection recall while maintaining high precision in scenes with heavy occlusions. Our NMS algorithm is derived from a novel embedding mechanism, in which the semantic and geometric features of the detected boxes are jointly exploited. The embedding makes it possible to determine whether two heavily-overlapping boxes belong to the same object in the physical world. Our approach is particularly useful for car detection and pedestrian detection in urban scenes where occlusions often happen. We show the effectiveness of our approach by creating a model called SG-Det (short for Semantics and Geometry Detection) and testing SG-Det on two widely-adopted datasets, KITTI and CityPersons for which it achieves state-of-the-art performance.

Acoustics Air-Sea Interaction & Remote Sensing Center for Environmental & Information Systems Center for Industrial & Medical Ultrasound Electronic & Photonic Systems Ocean Engineering Ocean Physics Polar Science Center
Close

 

Close