Method and apparatus for the detection of behaviours in a retail environment

Patent metadata

Field	Value
Publication number	US-12406503-B2
Application number	US-202218280136-A
Country	US
Kind code	B2
Filing date	Mar 3, 2022
Priority date	Mar 4, 2021
Publication date	Sep 2, 2025
Grant date	Sep 2, 2025

Abstract

A method and apparatus for the identification of suspect behaviour in a retail environment, the method comprising: detecting a person in a frame of said stream of video data; extracting a set of activities of the identified person from the stream of video data; assigning a numeric value to each extracted activity in the set of extracted activities, said numeric value representative of a threat level of the activity; accumulating said numeric values to provide a behaviour score; and identifying a behaviour as being suspect when the behaviour score reaches a target threshold value associated with the behaviour.

First claim

The invention claimed is:

1. A method for identification of suspect behaviour in a retail environment, comprising:
detecting a person in a frame of a stream of video data obtained from a plurality of video sensors in the retail environment, wherein detecting the person comprises establishing localization information for the detected person, by establishing a bounding box framing the person;
classifying the identified person as a tracked person or a non-tracked person;
tracking the path of the tracked person about the retail environment, wherein tracking the path of the tracked person comprises encoding the appearance of the person based on a plurality of semantic features selected from a list including visual appearance, body movement or interaction with the surroundings in the retail environment;
extracting by a behaviour detection unit, a set of activities of the tracked person from the one or more frames of the stream of video data, wherein extracting the set of activities comprises estimating a set of poses of tracked person, and wherein estimating the set of poses comprises identifying a predefined set of points on the tracked person and detecting successive movements of each of the predefined set of points over a time interval, and wherein the behaviour detection unit comprises a trajectory computation module adapted to output a predicted trajectory for the tracked person; an object detection module configured to detect an object which the tracked person picked up in the retail environment and assign a unique object identifier to the object; and a human pose estimation module for the detection of the set of activities or behaviours of the identified tracked person;
assigning a numeric value to each extracted activity in the set of extracted activities, said numeric value being representative of a threat level of the activity;
accumulating said numeric values to provide a behaviour score; and
identifying a behaviour as being suspect when the behaviour score reaches a target threshold value associated with the behaviour.

2. The method according to claim 1 further comprising assigning a unique identifier to the detected person.

3. The method according to claim 1 , wherein extracting the set of activities further comprises determining a vector representative of a trajectory of the identified person about the retail environment.

4. The method according to claim 1 wherein assigning the numeric value comprises assigning a value of 1 to each extracted activity and weighting the value based on a likelihood that the activity is associated with a behaviour to provide the numeric value.

5. The method according to claim 1 further comprising generating an alert when the accumulated behaviour score reaches the target threshold associated with the behaviour.

6. The method of claim 5 comprising setting a plurality of target threshold values, each target threshold value associated with a different alert from a set of predefined alerts.

7. The method according to claim 1 further comprising obtaining the stream of video data by a plurality of video sensors positioned to monitor selected zones of the retail environment.

8. A non-transitory computer readable storage medium carrying a computer program stored thereon which when executed by a processing module implements the method according to claim 1 .

9. An apparatus for identification of suspect client behaviour in a retail environment, comprising:
means for detecting a person in a frame of a stream of video data obtained from a plurality of video sensors in the retail environment, wherein detecting the person comprises establishing localization information for the detected person, by establishing a bounding box framing the person;
means for classifying the identified person as a tracked person or a non-tracked person;
means for tracking the path of the tracked person about the retail environment, wherein tracking the path of the tracked person comprises encoding the appearance of the person based on a plurality of semantic features selected from a list including visual appearance, body movement or interaction with the surroundings in the retail environment;
means for extracting by a behaviour detection unit, a set of activities of the tracked person from the one or more frames of the stream of video data, wherein extracting the set of activities comprises estimating a set of poses of tracked person, and wherein estimating the set of poses comprises identifying a predefined set of points on the tracked person and detecting successive movements of each of the predefined set of points over a time interval, and wherein the behaviour detection unit comprises a trajectory computation module adapted to output a predicted trajectory for the tracked person; an object detection module configured to detect an object which the tracked person picked up in the retail environment and assign a unique object identifier to the object; and a human pose estimation module for the detection of the set of activities or behaviours of the identified tracked person;
means for assigning a numeric value to each extracted activity in the set of extracted activities, said numeric value being representative of a threat level of the activity;
means for accumulating said numeric values to provide a behaviour score; and
means for identifying a behaviour as being suspect when the behaviour score reaches a target threshold value associated with the behaviour.

10. The apparatus of claim 9 wherein the means for detecting a person comprises a client tracking unit, means for extracting comprises a behaviour feature detection unit, means for assigning comprises a suspect activity detection unit; means for accumulating comprises a score update unit and means for identifying comprises a message issuing unit.

Assignees

• Everseen Ltd

Inventors

• Crisfalusi Dan Alin
• O'Herlihy Alan
• Todoran Cristina
• Gui Vasile
• Pescaru Dan
• David Ciprian Petru
• Cernazanu Cosmin
• Alexandru Arion

Classifications

• G06T2207/30241

  Trajectory · CPC title

• G06T7/20

  Analysis of motion (motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51) · CPC title

• G06V2201/07

  Target detection · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06V40/20

  Movements or behaviour, e.g. gesture recognition (recognition of facial expressions G06V40/16) · CPC title