VeV: Video-Enhanced Vibration Data Augmentation For Structural Difference

VeV: Video-Enhanced Vibration Data Augmentation For Structural Difference

The authors propose VeV, a framework for estimating vibration responses at locations without sensors by combining sparse vibration observations with visual information. They frame the problem as a spatial discontinuity: vibration patterns can change when sensors move or structures differ, while collecting new data and recalibrating can be costly. VeV uses a monocular Vision-Derived Propagation Possibility Map for structural wave-guiding paths and a Visual Feature Map for transient source kinetics and implicit material dynamics. In an evaluation on a 1,237 sensor-hour sow pig and piglet farm across disparate structures, the authors report waveform reconstruction MSE over 57.5% lower than data-driven baselines and up to 81.69% zero-shot classification accuracy. They also report peak performance within 30-50 calibration chunks, which they say can eliminate cold-start retraining overhead. The authors present human-centric floor vibration sensing and ambient intelligence as possible extensions; the abstract does not establish their performance in those settings.

Publication: 3ff28c42-565a-4c4a-89c6-ed61d334f8d9

attribution — pass: The reading attributes the framework, the problem framing and all performance figures to the authors ("The authors propose", "the authors report"). It does not present them as established facts. The author list and arXiv identity match the evidence.

Authors:Joonhee Lee, Jiale Zhang, Sungmin Lee, Ruiqing Wang, Jeffrey Shulkin, Yiwen Dong, Pei Zhang, JeongGil Ko

summary — uncertain: The reading is faithful but hard to follow without specialist background. It omits the stated purpose: non-invasive structural monitoring in settings such as smart farms. It also leaves central terms unexplained: MSE, zero-shot classification accuracy (and what is classified), calibration chunks, and cold-start retraining overhead. Smallest repair: add one plain sentence on what surface-mounted vibration sensing is used for. Also add or confirm term-library tooltips for MSE, zero-shot, calibration chunk, cold-start and monocular. The question is substantive and follows from the evidence.

Surface-mounted vibration sensing provides non-invasive structural monitoring with high temporal fidelity across diverse ubiquitous computing scenarios such as smart farms

date — pass: originalPublished is 2026-11-06, which matches the v1 submission date on arXiv. The feed date of Nov 10 is correctly not used.

[Submitted on 6 Nov 2026]

support — pass: The figures match the abstract exactly: over 57.5% MSE reduction, up to 81.69% zero-shot accuracy, 30-50 calibration chunks and 1,237 sensor-hours. The extension claim is appropriately labeled as interpretation and limited, consistent with the abstract's foundation language. The limitations correctly note that per-structure results and the evaluation protocol are absent. No unsupported benefits or costs are implied.

reaching peak performance within just 30-50 calibration chunks to effectively eliminate cold-start retraining overhead

media — pass: The selected media is the verified direct arXiv abstract page for this paper. It matches the source URL and is the most direct access to the source.