IRIS: A Real-World Benchmark for Inverse Recovery and Identification of Physical Dynamic Systems from Monocular Video
arXiv:2603.16432v3 Announce Type: replace-cross Abstract: Unsupervised physical parameter estimation from video lacks a common benchmark: existing methods evaluate on non-overlapping synthetic data, the sole real-world dataset is restricted to single-body systems, and no established protocol addresses governing-equation identification. This work introduces IRIS, a high-fidelity benchmark comprising 240 real-world videos captured at 4K resolution and 60fps, spanning both single- and multi-body dynamics with independently measured ground-truth parameters and uncertainty estimates. Each dynamical
The proliferation of high-resolution video capture and advances in AI for physical reasoning are converging, creating an opportune moment for robust real-world benchmarks.
This benchmark addresses a critical gap in AI research by providing a standardized, real-world dataset for understanding physical dynamic systems, crucial for areas like robotics and simulated environments.
The availability of IRIS will enable more rigorous and comparable evaluation of AI models designed for inverse physical recovery and system identification, moving beyond synthetic data limitations.
- · AI researchers (robotics, physics-based AI)
- · Robotics companies
- · Simulation software developers
- · Autonomous systems developers
- · AI methods reliant solely on synthetic or limited real-world data
Improved performance and generalization of AI models for understanding and predicting real-world physical interactions.
Accelerated development of more robust and adaptive robots capable of operating in complex, dynamic environments.
Potential for new AI-driven design and optimization tools for physical systems, leading to engineering innovation.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG