Efficient Visual Anomaly Detection at the Edge: Enabling Real-Time Industrial Inspection on Resource-Constrained Devices
Arianna Stropeni, Fabrizio Genilotti, Francesco Borsatti, Manuel Barusco, Davide Dalle Pezze, Gian Antonio Susto · 2026
Read Efficient Visual Anomaly Detection at the Edge: Enabling Real-Time Industrial Inspection on Resource-Constrained Devices by Arianna Stropeni, Fabrizio Genilotti, Francesco Borsatti, Manuel Barusco, Davide Dalle Pezze, Gian Antonio Susto (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-20288.txt.
Description
Stropeni and co-authors adapt visual anomaly detection for small edge devices in factories. Cloud processing clashes with real-time and privacy needs on production lines. They present PatchCore-Lite and PaDiM-Lite, lean versions of two known methods. PatchCore-Lite searches a compressed memory bank before checking a decoded subset. PaDiM-Lite uses diagonal covariance to speed its distance math. On MVTec AD and VisA, PatchCore-Lite cut memory use by 79%. PaDiM-Lite cut memory by 77% and run time by 31%.
Catalog record
| CREATOR | Arianna Stropeni, Fabrizio Genilotti, Francesco Borsatti, Manuel Barusco, Davide Dalle Pezze, Gian Antonio Susto |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 29505 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.20288