Granted Patent B2

US 12,548,371 · App. 18/107,766 · Granted Feb 10, 2026

Inventors: Hugo Latapie (Long Beach, CA); Gaowen Liu (Austin, TX); Ozkan Kilic (Long Beach, CA); Adam James Lawrence (Pasadena, CA); Ramana Rao V.R. Kompella (Cupertino, CA)

Assignee: Cisco Technology, Inc.

G06V40/20G06T7/73G06V10/70G06V20/41G06V20/46G06V20/52G06V40/10G06T2207/10016G06T2207/30196G06T2207/30232

Abstract

Claims (41)

1 . A method comprising:

representing, by a device, each of a plurality of objects depicted in video data over time as a set of timeseries of key points associated with that object;

forming, by the device, groups of objects by analyzing sets of timeseries of key points associate each object that are similar to one another;

identifying, by the device, an anomalous behavior of a particular group from among the groups of objects, based on their constituent timeseries of coordinate points and when compared to other groups; and

providing, by the device, an alert regarding the anomalous behavior to a user interface for review.

2 . The method as in claim 1 , further comprising:

detecting the key points by applying a pose estimation model to the video data.

3 . The method as in claim 1 , wherein the plurality of objects comprise people detected in the video data.

4 . The method as in claim 1 , wherein the groups of objects are formed based in part on frequencies or amplitudes of the set of timeseries.

5 . The method as in claim 1 , wherein the device uses self-supervised learning to identify the anomalous behavior.

6 . The method as in claim 1 , wherein the anomalous behavior comprises a set of timeseries associated with a particular object in one group moving to another group.

7 . The method as in claim 1 , further comprising:

using self-supervised learning to generate a label for the particular group, wherein the alert includes the label.

8 . The method as in claim 1 , wherein identifying the anomalous behavior comprises:

forming a hierarchy of the groups of objects whereby a level of the hierarchy comprises those of the groups of objects that have similar associated timeseries.

9 . The method as in claim 1 , wherein the groups of objects are formed based in part on spatial distances between the plurality of objects.

10 . The method as in claim 1 , wherein the device is an edge device in a network.

11 . An apparatus, comprising:

a network interface to communicate with a computer network;

a processor coupled to the network interface and configured to execute one or more processes; and

a memory configured to store a process that is executed by the processor, the process when executed configured to:

represent each of a plurality of objects depicted in video data over time as a set of timeseries of key points associated with that object;

form groups of objects by analyzing sets of timeseries of key points associated with each object that are similar to one another;

identify an anomalous behavior of a particular group from among the groups of objects, based on their constituent timeseries of coordinate points and when compared to other groups; and

provide an alert regarding the anomalous behavior to a user interface for review.

12 . The apparatus as in claim 11 , wherein the process when executed is further configured to:

detect the key points by applying a pose estimation model to the video data.

13 . The apparatus as in claim 11 , wherein the plurality of objects comprise people detected in the video data.

14 . The apparatus as in claim 11 , wherein the groups of objects are formed based in part on frequencies or amplitudes of the set of timeseries.

15 . The apparatus as in claim 11 , wherein the apparatus uses self-supervised learning to identify the anomalous behavior.

16 . The apparatus as in claim 11 , wherein the anomalous behavior comprises a set of timeseries associated with a particular object in one group moving to another group.

17 . The apparatus as in claim 11 , wherein the process when executed is further configured to:

use self-supervised learning to generate a label for the particular group, wherein the alert includes the label.

18 . The apparatus as in claim 11 , wherein the apparatus identifies the anomalous behavior by:

forming a hierarchy of the groups of objects whereby a level of the hierarchy comprises those of the groups of objects that have similar associated timeseries.

19 . The apparatus as in claim 11 , wherein the groups of objects are formed based in part on spatial distances between the plurality of objects.

20 . A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising:

representing, by the device, each of a plurality of objects depicted in video data over time as a set of timeseries of key points associated with that object;

forming, by the device, groups of objects by analyzing sets of timeseries of key points associated with each object that are similar to one another;

identifying, by the device, an anomalous behavior of a particular group from among the groups of objects, based on their constituent timeseries of coordinate points and when compared to other groups; and

providing, by the device, an alert regarding the anomalous behavior to a user interface for review.

Assignments (1)

Continuity (1)

References Cited (123)