معرفی
Gül Varol is a permanent researcher (equivalent to Associate Professor) at École des Ponts ParisTech, where she is part of the IMAGINE group within the Laboratoire d'Informatique Gaspard-Monge (LIGM). She holds additional affiliations as an ELLIS Scholar and a Guest Scientist at the Max Planck Institute (MPI). Her academic journey includes a postdoctoral position at the University of Oxford's Visual Geometry Group (VGG) under Andrew Zisserman, and a PhD from the WILLOW team at Inria Paris and École Normale Supérieure.
Her educational background includes BS and MS degrees from Boğaziçi University, followed by doctoral studies at Inria Paris and École Normale Supérieure. During her PhD, she spent research periods at MPI, Adobe, and Google, gaining diverse industry and academic experience.
Dr. Varol's research focuses on the intersection of computer vision and language processing, with particular expertise in video representation learning, human motion synthesis, and sign language technologies. Her work bridges theoretical advances with practical applications, especially in accessibility technologies like audio description for visually impaired audiences and sign language recognition systems. She has pioneered approaches for text-driven 3D human motion generation, sign language translation with contextual cues, and film-grammar-aware audio description generation. Her research often leverages large language models and vision-language models to solve complex multimodal problems without requiring extensive training data.
Analysis of her recent publications reveals a clear trajectory toward more sophisticated multimodal systems that integrate temporal understanding, contextual awareness, and fine-grained control. Her work increasingly focuses on practical applications in accessibility, with significant contributions to sign language technologies and audio description systems. She has demonstrated leadership in developing large-scale datasets like BOBSL (BBC-Oxford British Sign Language Dataset) and advancing methods for zero-shot and training-free approaches that can be deployed without extensive fine-tuning.
Her scientific achievements have been recognized with several prestigious awards:
- ELLIS 2020 PhD Award
- AFRIF 2019 PhD thesis award from the French association for pattern recognition
- Google Research Scholar award (2023)
- Best application paper award at ACCV'20
Dr. Varol has received significant research funding including an ANR JCJC project "CorVis" and research gifts from Google and Adobe. She actively mentors students and researchers, with prospective students encouraged to apply through her lab's application form. Her leadership in the academic community is evident through her service as Program Chair for ECCV'24, Senior Area Chair for CVPR 2025, and associate editor for the International Journal of Computer Vision (IJCV).
She is a key member of the IMAGINE research group at École des Ponts ParisTech, which focuses on image analysis and computer vision. Her collaborative work extends to multiple institutions including MPI, Oxford, and various industry research labs. She has co-organized numerous workshops on specialized topics including sign language recognition, vision transformers, and foundation models for 3D humans, fostering community building in these emerging research areas.
