UniMetric: Universal Metric Scale Recovery for Visual Geometry Foundation Models
Building on this insight, we present UniMetric, a universal and training-free metric scale recovery framework for geometry foundation models, without additional sensors or task-specific metric supervision.
ProofPaper ↗
Key points
- Recent advances in multimodal foundation models, exemplified by GPT-6 Astra, raise the prospect of general-purpose embodied reasoning and planning.
- Yet visual geometry foundation models often recover scene structure and camera motion without reliable metric scale.
- We show that uncertain size priors of familiar objects can jointly constrain a reconstruction's shared scale.
- To evaluate metric accuracy on GFM reconstructions without ground-truth scale alignment, we introduce MetricBench, covering twelve embodied, driving, and roadside subsets with sensor-derived ground truth.
Sources (1)
- [1]UniMetric: Universal Metric Scale Recovery for Visual Geometry Foundation ModelsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 12, 03:18 PM
Building on this insight, we present UniMetric, a universal and training-free metric scale recovery framework for geometry foundation models, without additional sensors or task-specific metric supervision.
Recent advances in multimodal foundation models, exemplified by GPT-6 Astra, raise the prospect of general-purpose embodied reasoning and planning.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 12, 2026Import AI 476: Creativity and RSI; worse-than-China open weight models; AI goes to space
- Oct 12, 2026Embodied Skill Bench: Benchmarking How Well Agents with Embodied Skills Perform on Physical Tasks
- Oct 12, 2026When Left Is Right? Camera-Motion Understanding beyond Screen
- Oct 11, 2026ReTrace: Cross-Stage Failure Attribution for Self-Improving Robotic Manipulation
- Oct 11, 2026Big-Big-SycoBench: A Benchmark to Detect Sycophancy Through Model Responses to Valid vs. Invalid Arguments