RelateAnything: Real-Time Open-Vocabulary Relation Prediction From Any Inputs
arXiv:2609.12552Maëlic Neau
I am the founder of an early-stage company in the Y Combinator Fall 2026 batch. We are pre-launch and not saying much publicly yet — more soon.
Before that I was a Computer Vision and Machine Learning Research Scientist, most recently a Postdoctoral Fellow at Umeå University, Sweden (May 2025 – 2026), working with Assoc. Prof. Zoe Falomir in the UMU Spatial Lab.
I hold a joint PhD (cotutelle) from ENIB, France and Flinders University, Australia (2025), with 5+ years of hands-on experience building real-time perception systems using object detection, Vision-Language Models, and autonomous robots with PyTorch, HuggingFace, ROS, etc.
I created RelateAnything, a real-time open-vocabulary relation prediction model that works with regions from any source and accepts the predicate vocabulary at inference. I am also the author of REACT (Oral at BMVC 2025) and SGG-Benchmark, the first open-source framework for real-time Scene Graph Generation.
I am a 2× World Champion of the RoboCup@Home competition (2022 & 2023) with the RoboBreizh team.
I am focused on the company right now, but always glad to hear from researchers and engineers working on real-time perception and embodied AI. Feel free to reach out!
Maëlic Neau
Maëlic Neau, Salim Baloch, Jakob Suchan, Zoe Falomir, Mehul Bhatt
Maëlic Neau, Zoe Falomir
Maëlic Neau, Zoe Falomir, Paulo E. Santos, Anne-Gwenn Bosser, Cédric Buche
Maëlic Neau, Paulo E. Santos, Anne-Gwenn Bosser, Akihiro Sugimoto, Cédric Buche
Maëlic Neau, Zoe Falomir, Cédric Buche, Akihiro Sugimoto
Maëlic Neau, Paulo E. Santos, Anne-Gwenn Bosser, Alistair Macvicar, Cédric Buche
Cédric Buche, Maëlic Neau, Thomas Ung, Louis Li, Sinuo Wang, Cédric Le Bono
Maëlic Neau, Paulo E. Santos, Anne-Gwenn Bosser and Cédric Buche
Sinuo Wang, Maëlic Neau and Cédric Buche
Maëlic Neau, Paulo Santos, Anne-Gwenn Bosser and Cédric Buche
Natnael Wondimu, Maëlic Neau, Antoine Dizet, Ubbo Visser and Cédric Buche
Louis Li, Maëlic Neau, Thomas Ung, and Cédric Buche
Cédric Buche, Maëlic Neau, Thomas Ung, Louis Li, Tianjao Jiang, Mukesh Barange and Maël Bouabdelli
Thanh Tran, Maëlic Neau, Paulo E. Santos and David Powers
Maëlic Neau, Paulo Santos, Anne-Gwenn Bosser, Cédric Buche and Nathan Beu
Download: CV (PDF)
Founder of an early-stage company in the Y Combinator Fall 2026 batch. Computer Vision & ML Research Scientist with a joint PhD and 5+ years building real-time perception systems, Vision-Language Models, and autonomous robots. Research experience across France, Australia, Japan, and Sweden.
ENIB, Brest - France / Flinders University, Adelaide - Australia
Topic: "Real-Time And Efficient Scene Graph Generation for Real-World Applications: An End-to-End Investigation"
Supervisors: Cédric Buche (CNRS / ENIB), Paulo E. Santos (Flinders University), Anne-Gwenn Bosser (ENIB) and Karl Sammut (Flinders University)
Université de Bretagne Occidentale, Brest - France / Umeå University, Umeå - Sweden
Artificial Intelligence and Robotics
Université de Bretagne Occidentale, Brest - France
IUT de Lannion, Lannion - France
Early-stage startup — Y Combinator, Fall 2026 batch
Computing Science Department, Umeå University, Sweden
National Institute of Informatics, Tokyo, Japan
Flinders University, South Australia
Lab-STICC / ENIB - Plouzané, France
Lab-STICC / ENIB - Plouzané, France
Northern Kentucky University - Highland Heights, USA