Selected Publications

In this paper we introduce the problem of saliency ranking, which arises from the observation that observers may not agree upon what is salient. We propose models, metrics and analysis associated with this problem.
In CVPR, 2018

In this paper we propose a gating mechanism that receives recurrent feedback from upstream feature activations. This produces significant improvements for recognition performance.
In CVPR, 2017

We consider visual saliency modeling including gaze prediction and salient object segmentation. This discourse reveals a number of high-level considerations relevant to deep-learning models for visual saliency prediction.
In CVPR, 2016

In this paper we present a definition for visual saliency grounded in information theory. This definition reveals strong ties between disparate concepts including scale-space theory, bilateral filtering, random walks and other phenomena.
In NIPS, 2015

Recent Publications

More Publications

. Gated Feedback Refinement Network for Dense Image Labeling. In CVPR, 2017.

PDF Project

. Gated feedback refinement network for dense image labeling. Proceedings - 30th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, 2017.

DOI

. Gaze-contingent interactive visualization of high-dynamic-range imagery. Proceedings of the 2nd Workshop on Eye Tracking and Visualization, ETVIS 2016, 2017.

DOI

. Movers, Shakers, and Those Who Stand Still: Visual Attention-grabbing Techniques in Robot Teleoperation. ACM/IEEE International Conference on Human-Robot Interaction, 2017.

DOI

. Shared Façades: Surface-embedded layout management for ad Hoc collaboration using head-worn displays. Collaboration Meets Interactive Spaces, 2017.

DOI

. Tortoise and the Hare Robot: Slow and steady almost wins the race, but finishes more safely. RO-MAN 2017 - 26th IEEE International Symposium on Robot and Human Interactive Communication, 2017.

DOI

. A deeper look at saliency: Feature contrast, semantics, and beyond. In CVPR, 2016.

. Dense image labeling using Deep Convolutional Neural Networks. Proceedings - 2016 13th Conference on Computer and Robot Vision, CRV 2016, 2016.

DOI

Projects

BCI and Neural Decoding

How can we build effective brain-computer interfaces and decode and understand neural activity?

Feedback and Gating

How can information be gated, selected and routed through deep neural networks?.

Selective Attention

How do we choose what to pay attention to, and in what order? How do machine vision systems?

Unsupervised Learning

How can we produce sophisticated AI without the need for labeled data?

Visual Saliency

What makes something salient? Models, metrics and other discussion.

Teaching

Courses for the current year (Ryerson University):

  • Winter 2019 - CPS 616 Advanced Algorithms
  • Fall 2018 - CPS 840 / 8318 Machine Learning
  • Fall 2018 - CP 8309 / 8315 Neural Information Processing

Previous courses (University of Manitoba):

  • Fall 2017 - COMP 7210 Research Methodologies
  • Fall 2017 - COMP 3490 Computer Graphics I
  • Winter 2017 - COMP 7210 Research Methodologies
  • Winter 2017 - COMP 4490 Computer Graphics II
  • Fall 2016 - COMP 7210 Research Methodologies
  • Fall 2016 - COMP 7960 Image Processing
  • Winter 2016 - COMP 4490 Computer Graphics II
  • Winter 2016 - COMP 7950 Computational Perception and Cognition
  • Winter 2015 - COMP 4490 Computer Graphics II
  • Winter 2015 - COMP 7210 Research Methodologies
  • Fall 2014 - COMP 4710 Data Mining
  • Fall 2014 - COMP 7210 Research Methodologies
  • Winter 2014 - COMP 4190 Artificial Intelligence
  • Winter 2014 - COMP 4490 Computer Graphics II
  • Fall 2013 - COMP 7960 Image Processing
  • Winter 2013 - COMP 4360 Machine Learning
  • Fall 2012 - COMP 7960 Image Processing

Opportunities

We welcome applications for an open postdoctoral position in the Ryerson Vision Lab.

CONTINUE READING