Here are a few of the many applications we focus on in this theme.
Anomaly detection in video
Context is important when looking for anomalies: A knife in a kitchen versus a classroom is different from normal behaviour, but how about during arts/crafts class?
Pipeline integrity monitoring
According to Natural Resources Canada, there are over 825,000km of pipelines in Canada. Together these pipeline networks deliver $106b worth of natural resources and 1.3billion barrels of oil transported annually. Pipeline accidents also have lasting impacts on the environment and public perception.
Thus, approaches which use variety of new sensing technology and AI for automated processing are needed to adhere to strict regulations and operate in a safe and efficient manner.
Data-driven modeling
We can learn models of the systems, understand the complex factors driving the system, and predict behaviours from the sensed data.
For example, fluid flow modeling is useful to monitor as flow properties is apparent in our everyday lives, natural systems, water basins (rivers, lakes), airflows.
Electricity grid modeling (power/electricity and their markets) is useful to understand and maintain efficient operations, while integrating impacts of new sustainable technologies such as renewable energy and microgrids.
Remote sensing
In remote sensing images which may be very large (i.e. several gigabytes, 30k x 30k pixels in a single acquisition), finding anomalies is akin to a needle in a haystack.
Detecting change between a scene imaged at different times is also subject to variability in the sensors, illumination, and objects.
