Revisor for Web

Product overview

Revisor is an AI-driven system built to observe and analyze electoral activity using neural networks and computer vision. It focuses on identifying and classifying events at polling locations with a high degree of reliability, making it useful for teams that need automated, continuous oversight during elections.

Main capabilities

• Detects and classifies voter actions and other behaviors captured on video feeds, helping separate legitimate voting from unrelated activity.
• Provides support for hand recounts and can flag footage for human review to speed up verification.
• Produces both immediate alerts and aggregated longer-term reports from recorded material.
• Can be adapted to different voting environments and procedural rules through supervised training.

Customization and deployment

Revisor’s models can be fine-tuned to reflect the rules and flow of varied electoral systems worldwide. Organizations can train the system on local procedures and camera setups so that its detections align with the unique requirements of a given country or jurisdiction.

Accuracy and validation

Field tests and controlled evaluations show the platform achieves accuracy levels as high as 98% in distinguishing voting events from other actions at polling stations. It has successfully identified anomalies in turnout figures and other discrepancies that suggest noncompliance with established electoral rules.

Practical uses

Revisor is particularly suited to election observation missions, where it serves as a scalable, lower-cost complement to traditional on-the-ground monitors. By automating portions of surveillance and analysis, it enhances oversight capacity and helps prioritize human intervention where it’s most needed.

Other options to consider

• Open-source election-monitoring projects and community-driven tools
• Subscription-based clients such as Canary Mail
• Large teams of in-person election observers and independent monitoring organizations