Submitted anonymously.

Submission data

Full name	Multi-CameraProbabilisticFusion
Description	A probabilistic multi-view reconstruction framework that fuses observations from multiple synchronized cameras and inertial measurements. The system estimates scene geometry using adaptive view weighting and uncertainty-aware depth aggregation. Camera poses are optimized jointly through a factor graph, resulting in robust localization and reconstruction under occlusions, motion blur, and varying illumination.
Parameters	Maximum optimization iterations: 120 View fusion radius: 0.15 m Depth uncertainty threshold: 0.05 Pose graph update frequency: 5 Hz Robust kernel scale: 1.2 IMU preintegration window: 20 frames
Programming language(s)	Python, PyTorch, CUDA
Hardware	A100
Submission creation date	23 Jun, 2026
Last edited	23 Jun, 2026

High-res multi-view results

Low-res many-view results

Low-res two-view results

SLAM results