Research topics

• Computational media intelligence: Media content is a source of rich, complex multimodal data (video, audio, metadata, text). Given the huge impact media has in our lives, content analysis can often help in insight generation, experience and interaction improvement. Our goal is to develop machine learning models to enable automated analysis of the audiovisual media content to understand the experience and impact of media and to discover societal trends through such analysis.

• Behavioral health analytics: Behavioral health is as important for our well-being as our physical health. Computational methods can provide nuanced analysis of behavioral health signals which is difficult to achieve by human inspection. This research thread aims at developing methodologies and algorithms for studying and monitoring mental health using physiological signals, video and audio.

• Behavior analysis for security: Analyzing and possibly predicting future behavior of an individual or a group/crowd can inform security and surveillance systems and ensure safety of autonomous vehicles. Our research along this direction includes pedestrian behavior understanding, crowd flow analysis and driver distraction monitoring.

PhD students

• Amir Shirian

• Atanu Samanta

• Daniel Catchpole

• Debaleena Roy

• Jiashu Liao

• Olayinka Ajayi

• Olly Styles

• Yiming Ma

Research support

• PI: Multimodal Learning for In-Car Driver Activity Monitoring. Ford University Research Program. 2021 - 2023.

• PI: Crossmodal Biometric Matching. Warwick RDF Award. 2019.

• PI: Application-aware Image Quality Assessment. Indian Space Research Organization. 2016 - 2018.

• CoI: Holistic Scene Understanding. Samsung Research. 2017 - 2018.

• PI: NVIDIA Academic GPU Grant. 2017.