Revisor is an AI-powered election monitoring system: a neural-network-based software package that learns to recognize physical objects, track their movements, and detect voting events from polling-station video records. It identifies ballot boxes, counts voters, and verifies turnout against official figures to surface discrepancies.
The system is trainable, allowing customization across different electoral systems and polling-station setups, and claims up to 98% accuracy in voter counting. Beyond detection, it supports formal complaint generation and violation identification, turning raw video into structured evidence rather than requiring manual review of every recording. In a notable real-world application, Revisor analyzed over 1 million hours of video from Russia’s 2021 Duma elections in partnership with independent outlet Meduza, identifying approximately 17 million potentially falsified ballots.
Revisor targets election observers, investigative journalists, and electoral authorities seeking large-scale polling station monitoring without deploying physical observers to every location. Pricing isn’t published on the site.
