Anna Rohrbach
Anna Rohrbach is a computer vision researcher whose work connects images and video with language. She earned an applied mathematics master's in 2010 and a computer science PhD at Saarland University in 2017, then became a postdoctoral researcher with Trevor Darrell at UC Berkeley. She coauthored work on movie descriptions, phrase grounding, video summarization, and multimodal claim verification; she received the German Pattern Recognition Award in 2023. Since 2023, she is a full professor of Multimodal Grounded Learning at TU Darmstadt and hessian.AI, co-leads the Multimodal AI Lab, and her coauthored VeriTaS benchmark won ACL's 2026 Best Resource Paper Award.
- Now at
- Multimodal AI Lab
Timeline
-
2010
An applied mathematics master's
In 2010, she earns an M.Sc. in Applied Mathematics at Odessa I.I.Mechnikov National University in Ukraine.
-
2015
A dataset for movie description
In 2015, she coauthors A Dataset for Movie Description, which aligns audio descriptions with full-length HD movies and includes aligned movie scripts. The MPII-MD corpus contains over 68,000 sentences and video snippets from 94 HD movies.
-
2015–2016
Grounding phrases in images
In 2015, she coauthors Grounding of Textual Phrases in Images by Reconstruction, which learns to locate text phrases by attending to image regions and reconstructing each phrase. The paper appears at ECCV 2016.
-
2017
A PhD and Berkeley postdoc
In 2017, she earns a PhD in Computer Science at Saarland University. By October 2017, she works as a postdoctoral researcher at UC Berkeley with Professor Trevor Darrell. In 2017, she coauthors Generating Descriptions with Grounded and Co-Referenced People, a movie-description model that grounds mentioned characters and resolves visual references between consecutive sentence or clip pairs.
-
2021
CLIP-It for video summaries
In 2021, she coauthors CLIP-It, a language-guided multimodal transformer for generic and query-focused video summarization that scores frames against a user query or an automatically generated video caption.
-
2023
A Darmstadt professorship and award
In 2023, she joins TU Darmstadt and hessian.AI as a full professor of Multimodal Grounded Learning, supported by a €2 million LOEWE Start Professorship from Hesse. She receives the German Pattern Recognition Award for her work at the intersection of vision and language.
-
2025
Conference roles and DEFAME
In 2025, she serves as an Area Chair for ICCV and joins the organizers' team as a Program Chair for ECCV 2026. She coauthors DEFAME, a zero-shot pipeline for open-domain text-image claim verification.
-
2026
VeriTaS and a TU professorship
In 2026, she coauthors VeriTaS, a dynamic benchmark for multimodal automated fact-checking that wins ACL's Best Resource Paper Award. Since 2023, she is a full professor of Multimodal Grounded Learning at TU Darmstadt and hessian.AI and co-leads the Multimodal AI Lab with Marcus Rohrbach.
Career
- Since 2023TU DarmstadtFull professor of Multimodal Grounded Learning
- Since 2023Multimodal AI LabCo-lead
- From 2017University of California, BerkeleyPostdoctoral researcher
Education
- 2017Saarland UniversityPhD in Computer Science
- 2010Odesa I. I. Mechnikov National UniversityM.Sc. in Applied Mathematics
Sources and credits
Photo: We have no published photo of Anna Rohrbach yet. Send us a free-licensed photo.
- Anna Rohrbach - Max Planck Institute for Informatics
- A Dataset for Movie Description
- Grounding of Textual Phrases in Images by Reconstruction
- Generating Descriptions with Grounded and Co-Referenced People
- CLIP-It! Language-Guided Video Summarization
- German Pattern Recognition Award 2023 presented to Prof. Anna Rohrbach for Groundbreaking Contributions to AI Research – Computer Science – TU Darmstadt
- Anna Rohrbach – Vision and Language
- DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts
- VeriTaS: The First Dynamic Benchmark for Multimodal Automated Fact-Checking
- Best Resource Paper Award for Fact-Checking Research at ACL 2026
- Overview – Multimodal AI – TU Darmstadt





