
Seminar: VLM-driven data and context curation for visual understanding

IEEE RO-MAN 2025

Events: PAISS – PRAIRIE/MIAI Artificial Intelligence Summer School
PanSt3R: Multi-view consistent panoptic segmentation
Learning to solve the multi-agent task assignment problem for automated data centers
Test-time vocabulary adaptation for language-driven object detection
Investigating robot behaviors to resolve navigation blocks

Efficient online text compression for RAG
Making RAG faster and leaner with new online compression techniques. OSCAR is the first fast and accurate soft online compression model for fine-tuned LLMs whilst PROVENCE, is a plug-and-play hard prompt model for any LLM.

NAVER LABS Europe @CVPR 2025
An overview of the 10 papers we're presenting at CVPR 2025. new 3D reconstruction modesl MUSt3R and Pow3R, visual localization, visual navigation, representation learning, semantic segmentation and human motion understanding.

Out of the box robot navigation and Spatial AI: Lessons learned moving AI out of simulation
Incorporating physics into embodied AI makes for smooth, fast and successful robot navigation in the real world.