Mohamed Mohamed (Mohamed Abdelpakey)

I'm currently a senior AI Engineer at Virtek Vision International and Adjunct Professor at the University of British Columbia.
Previously I was a postdoctoral researcher at
University of British Columbia - Okanagan, Computer Science department, where I worked on computer vision and machine learning. I obtained my PhD with distinction in Computer Engineering from Memorial University of Newfoundland where I worked on object tracking using deep learning.

My research interests are focused on Artificial Intelligence such as meta and online learning, object classification, tracking, segmentation, and detection. Recently, I have been working on self-supervised learning and meta-learning.
My research has been recognized by best-paper nomination at ISVC '19, my work with collaporators is highlighted on the front page of J. Imaging and UBCO Faculty of Science media release. Focusing on COVID testing based on automated analysis of xrays.

Email  /  CV  /  Google Scholar  /  Github

profile photo
Research

Here are samples of my work (I'm still adding my work for the complete list please go to this LINK).

Automated Human Cell Classification in Sparse Datasets using Few-Shot Learning
Reece Walsh, Mohamed H. Abdelpakey, Mohamed S Shehata
Subbmitted to Nature, 2021
arXiv

TransBlast: Self-Supervised Learning Using Augmented Subspace With Transformer for Background/Foreground Separation
Islam Osman, Mohamed H. Abdelpakey, Mohamed S Shehata
ICCVW, 2021
Code
DP-siam: Dynamic policy siamese network for robust object tracking
Mohamed H. Abdelpakey, Mohamed S. Shehata
IEEE Transactions on Image Processing, 2019
project page and code

The ninth visual object tracking vot2021 challenge results
Matej Kristan, Jiří Matas, ..., Mohamed H. Abdelpakey, Mohamed S. Shehata
ICCV, 2021
Code

Incorporating lidar and explicitly modeling the sky lets you reconstruct urban environments.

The sixth Visual Object Tracking VOT2018 challenge results
Matej Kristan, Aleˇs Leonardis, Jiří Matas, ..., Mohamed H. Abdelpakey, Mohamed S. Shehata
ICCV, 2021
Code

Dense depth completion techniques applied to freely-available sparse stereo data can improve NeRF reconstructions in low-data regimes.

Misc
Area Chair, CVPR 2022
Area Chair & Longuet-Higgins Award Committee Member, CVPR 2021
Area Chair, CVPR 2019
Area Chair, CVPR 2018

Design and source code from Jon Barron's website