XM4D: 4D Reconstruction through Certifiable Geometric Optimization
Overview
This research introduces XM4D, a framework for accurate and efficient 4D reconstruction that guarantees global optimality by separating camera motion estimation from non-rigid deformation into structured quadratically constrained quadratic programs (QCQP) solvable with semidefinite programming relaxations. The method lifts static point tracks to 3D using depth prediction models, reduces the problem to rotation-only QCQPs for both stages, and initializes dynamic Gaussian splatting for photometric refinement. Extensive experiments across seven benchmarks demonstrate accurate camera motion estimation on static and dynamic scenes, efficient optimization scaling to thousands of frames, and state-of-the-art novel-view rendering quality.
Generated from attributed reports · Updated 2 days ago
Event evidence and corrections
0 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
No current evidence-backed claims. Missing information remains not reported.
Report timeline
Follow attributed reports and material updates.
- arXiv Robotics — research abstractsSignalXM4D: 4D Reconstruction through Certifiable Geometric Optimization
This research introduces XM4D, a framework for accurate and efficient 4D reconstruction that guarantees global optimality by separating camera motion estimation from non-rigid deformation into structured quadratically constrained quadratic programs (QCQP) solvable with semidefinite programming relaxations. The method lifts static point tracks to 3D using depth prediction models, reduces the problem to rotation-only QCQPs for both stages, and initializes dynamic Gaussian splatting for photometric refinement. Extensive experiments across seven benchmarks demonstrate accurate camera motion estimation on static and dynamic scenes, efficient optimization scaling to thousands of frames, and state-of-the-art novel-view rendering quality.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).