Marc Pollefeys

UPAL achieves a remarkable 4x speedup and 10x smaller memory footprint while maintaining state-of-the-art performance in multi-view feature extraction.

GenRec achieves unparalleled reconstruction fidelity while enhancing perceptual quality in novel view synthesis by intelligently separating reconstruction from generation.

Monocular 3D object detection can achieve robust performance without the pitfalls of 2D-to-3D lifting, thanks to a novel integration of metric reconstruction and detection.

Jointly optimizing multi-room layouts as Manhattan polygons leads to significant gains in accuracy and robustness, outperforming traditional independent room estimation methods.

VidMap achieves unprecedented robustness and accuracy in metric reconstruction from uncalibrated videos, outperforming both SLAM and SfM methods under extreme conditions.

DVPSFormer reduces the computational burden of depth-aware video panoptic segmentation, enabling real-time decision-making for autonomous vehicles without sacrificing accuracy.

Achieving state-of-the-art hair dynamics in head avatars, DynHair allows for realistic and controllable hair movement that adapts to head motion.

LangLoc achieves unprecedented accuracy in indoor localization from natural language, closing the gap between coarse scene retrieval and precise pose estimation.

LIME turns ordinary egocentric video into a powerful tool for robots to dynamically adjust their camera poses based on user intent, revolutionizing how we think about robotic perception.

A minimalist approach to 3D reconstruction outperforms complex models, achieving sharper geometry and greater robustness in ambiguous scenarios.

SuperFlex achieves unprecedented reconstruction accuracy for 3D point clouds by enabling deformable superquadrics to represent complex geometries robustly.

Real-world robots can now leverage photo-realistic digital twins constructed directly from RGB-D videos, eliminating the need for manual modeling.

Achieving 60 FPS dynamic 4D hand reconstruction from egocentric videos, Hand-4DGS outperforms traditional methods by effectively handling occlusions and rapid motion.

Achieving state-of-the-art registration accuracy without any learned parameters or depth sensors, PROSE redefines the landscape of egocentric scene understanding.

GAM revolutionizes robot policy learning by seamlessly integrating 3D geometric reasoning, outperforming traditional models in accuracy and efficiency.

A compact scene graph representation allows for efficient visual localization, slashing storage needs while preserving accuracy in complex environments.

Achieving six times fewer Gaussians while surpassing state-of-the-art performance redefines efficiency in 3D scene reconstruction.

Finally, a feed-forward 3D reconstruction method that spits out meshes ready for physics engines, no expensive post-processing needed.

Classical SfM can get stuck, and feedforward reconstruction can be brittle, but combining them creates a system that's both robust and accurate.

Finally, a feed-forward method cracks dynamic 3D scene reconstruction from multi-view video without needing camera poses, opening the door to real-time applications.

Forget training wheels: this training-free method leverages uncertainty to guide vision-language models to the right image regions, boosting performance on detail-oriented tasks.