Event detection based on video metadata

摘要

Video data from sources, such as cameras, is analyzed to create metadata descriptive of the events and objects occurring in the video. This metadata, which consists of data about the video, is then analyzed on a transaction basis to determine if a suspicious activity, such as a fraudulent Point of Sale (POS) return event, has occurred in relation to a transaction.

主权项

1. A system for identifying possible fraudulent activity in connection with a transaction which occurred at point-of-sale (POS) device, an employee region of interest and a customer region of interest being associated with the POS device, the system comprising at least one or more programmed processors and one or more storage devices cooperating to: identify video metadata corresponding to activity which took place in the customer and employee regions of interest during a time period related to the transaction; based on the examination of the metadata, being configured to: determine whether a customer was located in the customer region of interest during the time period, the determination including: (i) identifying a set of objects, if any, in the customer region of interest which meet predetermined criteria; and (ii) excluding from that set of objects one or more objects as a function of whether they are also located in the employee region of interest; wherein the excluding includes: determining a percentage of surface area overlap, of the one or more objects of interest from the set of objects, between the customer region of interest and the employee region of interest and using the percentage information to determine in which region the one or more objects is located.