Suman Saha

Suman pursued his doctoral studies in the Visual Artificial Intelligence Laboratory at Oxford Brookes University, United Kingdom, focusing on spatiotemporal human action localization using deep learning techniques. Suman received a Ph.D. in Computer Science and Mathematics from Oxford Brookes University in 2017. He served as a postdoctoral fellow at Oxford Brookes University between December 2017 and July 2018. Then he moved to CVL (Computer Vision Lab) at ETH Zurich, holding a postdoctoral researcher position until June 2023. Suman's research has centered around unsupervised domain adaptation (UDA) for visual scene understanding (semantic and panoptic segmentation), human behavior understanding, and vision-based biometrics (face anti-spoofing). He also worked on semi-supervised learning for semantic segmentation by leveraging self-supervised depth estimation. His internship at Disney Research Zurich involved designing deep generative models for unsupervised facial expression learning. Additionally, Suman tackled research problems in multi-task learning (MTL) by addressing two common challenges in developing multi-task models, incremental learning and task interference.