Mikhail Kennerley

Hello! I am a PhD Candidate working on high-level Computer Vision at the National University of Singapore (ECE) under the supervision of Prof. Robby Tan and Prof. Bharadwaj Veeravalli. I’m also affliated with I2R, A*STAR under the supervision of Dr. Wang Jian-Gang. My research focuses on domain adaptation for adverse weather conditions and semi-supervised learning. Currently doing an internship at HTX Sense-making & Surveillance.

Outside of research, I lead a series of career-based workshops in Mendaki Club for youths in the Malay/Muslim community.

Don’t hesitate to get in touch!

Email  /  Google Scholar  /  Twitter  /  Github

profile photo
News
  • June 2024: Staring my internship in HTX Sense-making & Surveillance.
  • Feb 2024: Paper on class-imbalance in domain adaptive object detection accepted into CVPR'24!
  • Nov 2023: Invited to be part of a Youth Panel to work with MCI on developing policies on Digital Well-being.
  • Aug 2023: Participated in Our Singapore Leadership Programme by the National Youth Council (NYC).
  • Feb 2023: First paper accepted into CVPR!
Research

My current research revolves around domain adaptation and imbalanced classes for object detection. I am also interested in generative datasets to aid in domain generalisation.

CAT: Exploiting Inter-Class Dynamics for Domain Adaptive Object Detection
Mikhail Kennerley, Wang Jian-Gang, Bharadwaj Veeravalli, Robby T. Tan
CVPR, 2024
project page / arXiv

Using inter-class relations to address the class-imbalance problem in domain adaptive object detection. Code soon to be released.

2PCNet: Two-Phase Consistency Training for Day-to-Night Unsupervised Domain Adaptive Object Detection
Mikhail Kennerley, Wang Jian-Gang, Bharadwaj Veeravalli, Robby T. Tan
CVPR, 2023
project page / arXiv / code

Utalising low-confidence samples in the student-teacher network to learn more of the target domain.

Misc
NUS Teaching Assistant, EE4701
Teaching Assistant, CEG5304
Teaching Assistant, EE6934
Teaching Assistant, EE5731

Website template stolen from Jon Barron.