Gül Varol

Gül Varol is a permanent researcher at École des Ponts ParisTech's IMAGINE group, an ELLIS Scholar, and Guest Scientist at Max Planck Institute. She holds a PhD from Inria Paris/ENS with awards from ELLIS and AFRIF. Her academic service includes Program Chair at ECCV'24 and Area Chair roles at major conferences.

• Current affiliations: IMAGINE group (École des Ponts ParisTech), Max Planck Institute
• Previous roles: Postdoctoral researcher at University of Oxford

Her research focuses on vision-language applications, particularly in 3D human motion synthesis, sign language technology, and audio description generation. Key techniques include text-conditioned diffusion models, temporal context modeling, and synthetic data utilization.

Scientific contributions recognized through:

• Google Research Scholar award (2023)
• ELLIS PhD Award (2020)
• AFRIF PhD thesis award (2020)
• Best application paper at ACCV'20

Recent publications demonstrate expertise in:

• Text-driven 3D motion editing (MotionFix, 2024)
• Cross-dataset generalization studies (TMR++, 2024)
• Temporal action composition frameworks (TEACH, 2022)
• Sign language dense annotation methods (BOBSL, 2022)
• Zero-shot audio description generation (AutoAD-Zero, 2024)

She actively contributes to dataset development including BOBSL (British Sign Language corpus) and SURREACT synthetic action dataset, while pioneering new evaluation metrics for audio description quality and motion retrieval benchmarks.