Lab Research Summary

My research group focuses on detecting, classifying, and tracking objects or people in both images and video. Our research aims also include special consideration of settings including out-of-domain or sparsely labeled samples. Application of our work to drones is a specific end-goal for our research group. More precisely, we currently target the following fields:

• Object Tracking: Our past work in this field has involved creation of novel object tracking approaches that are capable of continuously following objects or people in challenging scenarios. Re-identification of targets in aerial imagery is one notable application of our work.
• Few-Shot Learning: Our previous work in few-shot learning has enabled state-of-the-art object classification using small amounts of labelled data on in-domain or out-of-domain settings. Classification of out-of-domain aerial or satellite landscapes is another significant application of our work.
• Object Segmentation: Our research group members have proposed novel segmentation networks that address difficult settings for object and people-based segmentation.
• Self-Supervised Learning: Our goals include application of novel pretext tasks and use of unlabelled data to improve the out-of-domain robustness and overall accuracy of our techniques.

News:

Nov 14, 2022: Congratulations Dr. Shehata for being the recipient of the 2022 IEEE Canada Presidents’ make-a-difference award, “For significant contributions to IEEE Canada and to advancing the science and engineering profession”

Nov 4, 2022: Congratulations Dr. Shehata and the whole research group for being listed on the 2022 Stanford University Top 2% Scientists that was released on Nov 2, 2022 in the specialization of artificial intelligence and image processing. This is great as this is the second year in a row to be on the list since it was launched in 2019