Hazel Doughty

Hazel Doughty is an Assistant Professor at Leiden University in the Leiden Institute for Advanced Computer Science (LIACS). Previously, she was a postdoctoral researcher at the University of Amsterdam and completed her PhD at the University of Bristol under the supervision of Prof. Dima Damen and Prof. Walterio Mayol-Cuevas.

• Research Interests: Video Understanding, Skill Determination, Self-Supervised Learning, Temporal Attention, and Adverb Recognition in Instructional Videos.
• Grants: Co-Applicant for DUAL-IMPACT (NWO High Tech Systems and Materials, €1.25M); Main Applicant for NWO Veni grant (From What to How: Perceiving Subtle Differences in Videos, €280K).
• Scientific Recognition:
  • Veni Grant (2023)
  • ELLIS Member (2022)
  • Outstanding Reviewer for CVPR, NeurIPS, ECCV, and ACCV
• Academic Service: Organizer of workshops at CVPR 2024, BMVC 2023, NCCV 2024, and NeurIPS'21. Area Chair for CVPR, ICCV, NeurIPS, AAAI.
• Teaching: Courses on Computer Vision (BSc) and Advances in Deep Learning (MSc) at Leiden University.

Her work focuses on fine-grained video understanding with weak or incomplete supervision, including adverb analysis, self-supervised learning, and egocentric vision datasets. Her recent publications analyze benchmark sensitivity, motion-focused video-language models, and generalized category discovery.

Collaborations include the HAVA lab at the University of Amsterdam and the EPIC-Kitchens-100 project. She leads the DUAL-IMPACT initiative on high-tech systems and co-organized workshops at CVPR, ICCV, and NeurIPS. Her students include PhD candidates Luc Sträter and Kaiting Liu, and former advisees like Fida Mohammad Thoker (now postdoc at KAUST) and Piyush Bagad (now PhD at Oxford).