Fabian Deuser

About Me

I am a Ph.D. student at the University of the Bundeswehr Munich and TUM, with a primary research focus on cross-view geo-localisation. Currently, I am delving into how this domain can benefit from advancements in other areas. Specifically, I am exploring the synergies between cross-view geo-localisation and multimodal deep learning, as well as the potential enhancements image synthesis can bring to the field.

If you want to collaborate just write me a mail.

Research Interests

• Computer Vision: image retrieval, cross-view image matching, image synthesis
• Multimodal Deep Learning: multimodal representations, large multimodal models
• Machine Learning: contrastive learning, representation learning

News

• [Aug. 2023] Our paper about cross-view geo-localisation for drones is accepted to the ACM MM 23 UAVM Workshop. We also won first place in the corresponding competition.
• [Mar. 2023] Our paper investigating hard negative sampling for cross-view matching is accepted to the ICCV 2023.

Selected Publications

Awards and Competitions

• 1st Place ACM MM 2023 - UAVs in Multimedia: Capturing the World from a New Perspective Challenge [Page]
• 1st Place CVPR 2023 SoccerNet Challenge - Player Re-Identification Challenge [Page]
• Jury Award CVPR 2023 SoccerNet Challenge - Most Innovative Runner-Up Method [Page]
• 2nd Place CVPR 2023 SoccerNet Challenge - Player Jersey Number Recognition [Page]
• 1st Place ACM MM 2022 - Workshop on Multimedia Content Analysis in Sports Player Re-Identification Challenge [Page]
• Outstanding Reviewer ACM MM 2023 UAVM Workshop [Page]

Services

• Workshop Reviewer:
  • ACM MM 2023 UAVM Workshop
  • ACM SigSpatial 2023 GeoSearch Workshop
• Program Committee Member:
  • ACM SigSpatial 2023 GeoSearch Workshop [Page]