Verfahren und Systeme zur Zusammenfassung von Ereignissen, die von einem Videoüberwachungssystem Erfasst Werden und Zugehörige Alarme Bereitstellen

Details

Veröffentlichungs-Nr.

EP4779966

Anmeldetag

2. Januar 2026

Veröffentlichung

22. Juli 2026

Rechtsraum

EP

IPC

H04N7/18

Abstract

A video surveillance system and method that uses artificial intelligence to process video streams from multiple cameras. A Generative AI video-to-text model generates text summaries of events from each camera's video stream. A Large Language Model (LLM) processes these text summaries to create a cumulative summary of events. In some cases, the system can receive audio from microphones associated with cameras and convert it to text summaries using an audio-to-text model. The LLM can generate real-time text alerts with details from video and audio streams, which can be converted to event tailored audio alerts using text-to-speech technology. The system may provide operators with access to the text summaries and allows keyword searching of events with links to corresponding video and/or audio clips. The LLM can aggregate multiple related events into unified event entries with descriptive summaries.

Anmelder

Firma

Honeywell International Inc.

Land

🇺🇸 USA

Fachgebiete