Automatic event detection, text generation, and use thereof

摘要

A system and method for video surveillance and searching are disclosed. Video is analyzed and events are automatically detected. Based on the automatically detected events, textual descriptions are generated. The textual descriptions may be used to supplement video viewing and event viewing, and to provide for textual searching for events.

主权项

1. A video surveillance system, comprising:
at least a first video camera configured to capture at least a first video sequence including a plurality of video images; a processing system configured to automatically generate video image descriptive data based on the video images, and, based on the automatically generated video image descriptive data, to automatically detect events that involve at least one agent in the video sequence; a storage system configured to receive and store records of the automatically detected events and to associate the automatically detected events with at least a time associated with the first video sequence; and a text generator configured to automatically generate textual descriptions of the automatically detected events, wherein the storage system is further configured to store the automatically detected events with respective geographical locations in association with each other, the associated geographical location being at least in part based on GPS, Geographic Information Systems (GIS), or map data, and wherein the detected events are events that relate to map data and are based on rules that relate to the map data, such that the text generator generates text that combines the map data with the detected events into a report.